c-api.h File Reference

Public C API for sherpa-onnx. More...

#include <stdint.h>

Include dependency graph for c-api.h:

This graph shows which files directly or indirectly include this file:

Go to the source code of this file.

Classes
struct	SherpaOnnxOnlineTransducerModelConfig
	Configuration for a streaming transducer model. More...
struct	SherpaOnnxOnlineParaformerModelConfig
	Configuration for a streaming Paraformer model. More...
struct	SherpaOnnxOnlineZipformer2CtcModelConfig
	Configuration for a streaming Zipformer2 CTC model. More...
struct	SherpaOnnxOnlineNemoCtcModelConfig
	Configuration for a streaming NeMo CTC model. More...
struct	SherpaOnnxOnlineToneCtcModelConfig
	Configuration for a streaming T-One CTC model. More...
struct	SherpaOnnxOnlineModelConfig
	Model configuration shared by streaming ASR recognizers. More...
struct	SherpaOnnxFeatureConfig
	Feature extraction settings for ASR. More...
struct	SherpaOnnxOnlineCtcFstDecoderConfig
	Configuration for HLG/FST-based online CTC decoding. More...
struct	SherpaOnnxHomophoneReplacerConfig
	Configuration for homophone replacement. More...
struct	SherpaOnnxOnlineRecognizerConfig
	Configuration for a streaming ASR recognizer. More...
struct	SherpaOnnxOnlineRecognizerResult
	Incremental recognition result for a streaming ASR stream. More...
struct	SherpaOnnxOfflineTransducerModelConfig
	Configuration for a non-streaming transducer model. More...
struct	SherpaOnnxOfflineParaformerModelConfig
	Configuration for a non-streaming Paraformer model. More...
struct	SherpaOnnxOfflineNemoEncDecCtcModelConfig
	Configuration for a non-streaming NeMo CTC model. More...
struct	SherpaOnnxOfflineWhisperModelConfig
	Configuration for a non-streaming Whisper model. More...
struct	SherpaOnnxOfflineCanaryModelConfig
	Configuration for a Canary model. More...
struct	SherpaOnnxOfflineCohereTranscribeModelConfig
	Configuration for a Cohere Transcribe model. More...
struct	SherpaOnnxOfflineFireRedAsrModelConfig
	Configuration for a FireRedAsr encoder/decoder model. More...
struct	SherpaOnnxOfflineFireRedAsrCtcModelConfig
	Configuration for a FireRedAsr CTC model. More...
struct	SherpaOnnxOfflineMoonshineModelConfig
	Configuration for a Moonshine model. More...
struct	SherpaOnnxOfflineTdnnModelConfig
	Configuration for a TDNN model. More...
struct	SherpaOnnxOfflineLMConfig
	Configuration for an offline language model. More...
struct	SherpaOnnxOfflineSenseVoiceModelConfig
	Configuration for a SenseVoice model. More...
struct	SherpaOnnxOfflineDolphinModelConfig
	Configuration for a Dolphin model. More...
struct	SherpaOnnxOfflineZipformerCtcModelConfig
	Configuration for an offline Zipformer CTC model. More...
struct	SherpaOnnxOfflineWenetCtcModelConfig
	Configuration for an offline WeNet CTC model. More...
struct	SherpaOnnxOfflineOmnilingualAsrCtcModelConfig
	Configuration for an omnilingual offline CTC model. More...
struct	SherpaOnnxOfflineFunASRNanoModelConfig
	Configuration for an offline FunASR Nano model. More...
struct	SherpaOnnxOfflineQwen3ASRModelConfig
	Configuration for an offline Qwen3-ASR model. More...
struct	SherpaOnnxOfflineMedAsrCtcModelConfig
	Configuration for a MedASR CTC model. More...
struct	SherpaOnnxOfflineModelConfig
	Model configuration shared by offline ASR recognizers. More...
struct	SherpaOnnxOfflineRecognizerConfig
	Configuration for a non-streaming ASR recognizer. More...
struct	SherpaOnnxOfflineRecognizerResult
	Recognition result for a non-streaming ASR stream. More...
struct	SherpaOnnxKeywordResult
	Snapshot of the current keyword spotting result. More...
struct	SherpaOnnxKeywordSpotterConfig
	Configuration for keyword spotting. More...
struct	SherpaOnnxSileroVadModelConfig
	Configuration for a Silero VAD model. More...
struct	SherpaOnnxTenVadModelConfig
	Configuration for a Ten VAD model. More...
struct	SherpaOnnxVadModelConfig
	Configuration shared by voice activity detectors. More...
struct	SherpaOnnxSpeechSegment
	One detected speech segment returned by the VAD. More...
struct	SherpaOnnxOfflineTtsVitsModelConfig
	Configuration for a VITS TTS model. More...
struct	SherpaOnnxOfflineTtsMatchaModelConfig
	Configuration for a Matcha TTS model. More...
struct	SherpaOnnxOfflineTtsKokoroModelConfig
	Configuration for a Kokoro TTS model. More...
struct	SherpaOnnxOfflineTtsKittenModelConfig
	Configuration for a Kitten TTS model. More...
struct	SherpaOnnxOfflineTtsZipvoiceModelConfig
	Configuration for a ZipVoice TTS model. More...
struct	SherpaOnnxOfflineTtsPocketModelConfig
	Configuration for a Pocket TTS model. More...
struct	SherpaOnnxOfflineTtsSupertonicModelConfig
	Configuration for a Supertonic TTS model. More...
struct	SherpaOnnxOfflineTtsModelConfig
	Configuration shared by offline TTS models. More...
struct	SherpaOnnxOfflineTtsConfig
	Configuration for offline text-to-speech. More...
struct	SherpaOnnxGeneratedAudio
	Generated waveform returned by TTS APIs. More...
struct	SherpaOnnxGenerationConfig
	Generation-time parameters shared by advanced TTS APIs. More...
struct	SherpaOnnxWave
	Decoded mono WAVE file content. More...
struct	SherpaOnnxMultiChannelWave
	Decoded multi-channel WAVE file content. More...
struct	SherpaOnnxSpokenLanguageIdentificationWhisperConfig
	Whisper-based model files for spoken language identification. More...
struct	SherpaOnnxSpokenLanguageIdentificationConfig
	Configuration for spoken language identification. More...
struct	SherpaOnnxSpokenLanguageIdentificationResult
	Result of spoken-language identification. More...
struct	SherpaOnnxSpeakerEmbeddingExtractorConfig
	Configuration for speaker embedding extraction. More...
struct	SherpaOnnxSpeakerEmbeddingManagerSpeakerMatch
	One speaker match returned by the best-matches API. More...
struct	SherpaOnnxSpeakerEmbeddingManagerBestMatchesResult
	Collection of best speaker matches. More...
struct	SherpaOnnxOfflineZipformerAudioTaggingModelConfig
	Zipformer audio-tagging model configuration. More...
struct	SherpaOnnxAudioTaggingModelConfig
	Audio-tagging model configuration. More...
struct	SherpaOnnxAudioTaggingConfig
	Configuration for audio tagging. More...
struct	SherpaOnnxAudioEvent
	One audio-tagging prediction. More...
struct	SherpaOnnxOfflinePunctuationModelConfig
	Offline punctuation model configuration. More...
struct	SherpaOnnxOfflinePunctuationConfig
	Configuration for offline punctuation. More...
struct	SherpaOnnxOnlinePunctuationModelConfig
	Online punctuation model configuration. More...
struct	SherpaOnnxOnlinePunctuationConfig
	Configuration for online punctuation. More...
struct	SherpaOnnxResampleOut
	Output chunk returned by SherpaOnnxLinearResamplerResample(). More...
struct	SherpaOnnxOfflineSpeakerSegmentationPyannoteModelConfig
	Pyannote speaker-segmentation model configuration. More...
struct	SherpaOnnxOfflineSpeakerSegmentationModelConfig
	Segmentation model configuration for offline speaker diarization. More...
struct	SherpaOnnxFastClusteringConfig
	Fast clustering configuration. More...
struct	SherpaOnnxOfflineSpeakerDiarizationConfig
	Configuration for offline speaker diarization. More...
struct	SherpaOnnxOfflineSpeakerDiarizationSegment
	One diarization segment. More...
struct	SherpaOnnxOfflineSpeechDenoiserGtcrnModelConfig
	GTCRN offline denoiser model configuration. More...
struct	SherpaOnnxOfflineSpeechDenoiserDpdfNetModelConfig
	DPDFNet offline denoiser model configuration. More...
struct	SherpaOnnxOfflineSpeechDenoiserModelConfig
	Speech denoiser model configuration shared by offline and online APIs. More...
struct	SherpaOnnxOfflineSpeechDenoiserConfig
	Configuration for offline speech denoising. More...
struct	SherpaOnnxDenoisedAudio
	Denoised audio returned by offline or online speech enhancement APIs. More...
struct	SherpaOnnxOnlineSpeechDenoiserConfig
	Configuration for streaming speech denoising. More...
struct	SherpaOnnxOfflineSourceSeparationSpleeterModelConfig
	Spleeter source-separation model configuration. More...
struct	SherpaOnnxOfflineSourceSeparationUvrModelConfig
	UVR (MDX-Net) source-separation model configuration. More...
struct	SherpaOnnxOfflineSourceSeparationModelConfig
	Source-separation model configuration. More...
struct	SherpaOnnxOfflineSourceSeparationConfig
	Top-level source-separation configuration. More...
struct	SherpaOnnxSourceSeparationStem
	A single stem (one output track) with one or more channels. More...
struct	SherpaOnnxSourceSeparationOutput
	Output of a source-separation run. More...

Macros
#define	SHERPA_ONNX_EXPORT   __attribute__((visibility("default")))
#define	SHERPA_ONNX_IMPORT   SHERPA_ONNX_EXPORT
#define	SHERPA_ONNX_API   SHERPA_ONNX_IMPORT

Typedefs
typedef struct SherpaOnnxOnlineTransducerModelConfig	SherpaOnnxOnlineTransducerModelConfig
	Configuration for a streaming transducer model.
typedef struct SherpaOnnxOnlineParaformerModelConfig	SherpaOnnxOnlineParaformerModelConfig
	Configuration for a streaming Paraformer model.
typedef struct SherpaOnnxOnlineZipformer2CtcModelConfig	SherpaOnnxOnlineZipformer2CtcModelConfig
	Configuration for a streaming Zipformer2 CTC model.
typedef struct SherpaOnnxOnlineNemoCtcModelConfig	SherpaOnnxOnlineNemoCtcModelConfig
	Configuration for a streaming NeMo CTC model.
typedef struct SherpaOnnxOnlineToneCtcModelConfig	SherpaOnnxOnlineToneCtcModelConfig
	Configuration for a streaming T-One CTC model.
typedef struct SherpaOnnxOnlineModelConfig	SherpaOnnxOnlineModelConfig
	Model configuration shared by streaming ASR recognizers.
typedef struct SherpaOnnxFeatureConfig	SherpaOnnxFeatureConfig
	Feature extraction settings for ASR.
typedef struct SherpaOnnxOnlineCtcFstDecoderConfig	SherpaOnnxOnlineCtcFstDecoderConfig
	Configuration for HLG/FST-based online CTC decoding.
typedef struct SherpaOnnxHomophoneReplacerConfig	SherpaOnnxHomophoneReplacerConfig
	Configuration for homophone replacement.
typedef struct SherpaOnnxOnlineRecognizerConfig	SherpaOnnxOnlineRecognizerConfig
	Configuration for a streaming ASR recognizer.
typedef struct SherpaOnnxOnlineRecognizerResult	SherpaOnnxOnlineRecognizerResult
	Incremental recognition result for a streaming ASR stream.
typedef struct SherpaOnnxOnlineRecognizer	SherpaOnnxOnlineRecognizer
	Streaming recognizer handle.
typedef struct SherpaOnnxOnlineStream	SherpaOnnxOnlineStream
	Streaming decoding state for one utterance or stream.
typedef struct SherpaOnnxDisplay	SherpaOnnxDisplay
	Helper for pretty-printing incremental recognition results.
typedef struct SherpaOnnxOfflineTransducerModelConfig	SherpaOnnxOfflineTransducerModelConfig
	Configuration for a non-streaming transducer model.
typedef struct SherpaOnnxOfflineParaformerModelConfig	SherpaOnnxOfflineParaformerModelConfig
	Configuration for a non-streaming Paraformer model.
typedef struct SherpaOnnxOfflineNemoEncDecCtcModelConfig	SherpaOnnxOfflineNemoEncDecCtcModelConfig
	Configuration for a non-streaming NeMo CTC model.
typedef struct SherpaOnnxOfflineWhisperModelConfig	SherpaOnnxOfflineWhisperModelConfig
	Configuration for a non-streaming Whisper model.
typedef struct SherpaOnnxOfflineCanaryModelConfig	SherpaOnnxOfflineCanaryModelConfig
	Configuration for a Canary model.
typedef struct SherpaOnnxOfflineCohereTranscribeModelConfig	SherpaOnnxOfflineCohereTranscribeModelConfig
	Configuration for a Cohere Transcribe model.
typedef struct SherpaOnnxOfflineFireRedAsrModelConfig	SherpaOnnxOfflineFireRedAsrModelConfig
	Configuration for a FireRedAsr encoder/decoder model.
typedef struct SherpaOnnxOfflineFireRedAsrCtcModelConfig	SherpaOnnxOfflineFireRedAsrCtcModelConfig
	Configuration for a FireRedAsr CTC model.
typedef struct SherpaOnnxOfflineMoonshineModelConfig	SherpaOnnxOfflineMoonshineModelConfig
	Configuration for a Moonshine model.
typedef struct SherpaOnnxOfflineTdnnModelConfig	SherpaOnnxOfflineTdnnModelConfig
	Configuration for a TDNN model.
typedef struct SherpaOnnxOfflineLMConfig	SherpaOnnxOfflineLMConfig
	Configuration for an offline language model.
typedef struct SherpaOnnxOfflineSenseVoiceModelConfig	SherpaOnnxOfflineSenseVoiceModelConfig
	Configuration for a SenseVoice model.
typedef struct SherpaOnnxOfflineDolphinModelConfig	SherpaOnnxOfflineDolphinModelConfig
	Configuration for a Dolphin model.
typedef struct SherpaOnnxOfflineZipformerCtcModelConfig	SherpaOnnxOfflineZipformerCtcModelConfig
	Configuration for an offline Zipformer CTC model.
typedef struct SherpaOnnxOfflineWenetCtcModelConfig	SherpaOnnxOfflineWenetCtcModelConfig
	Configuration for an offline WeNet CTC model.
typedef struct SherpaOnnxOfflineOmnilingualAsrCtcModelConfig	SherpaOnnxOfflineOmnilingualAsrCtcModelConfig
	Configuration for an omnilingual offline CTC model.
typedef struct SherpaOnnxOfflineFunASRNanoModelConfig	SherpaOnnxOfflineFunASRNanoModelConfig
	Configuration for an offline FunASR Nano model.
typedef struct SherpaOnnxOfflineQwen3ASRModelConfig	SherpaOnnxOfflineQwen3ASRModelConfig
	Configuration for an offline Qwen3-ASR model.
typedef struct SherpaOnnxOfflineMedAsrCtcModelConfig	SherpaOnnxOfflineMedAsrCtcModelConfig
	Configuration for a MedASR CTC model.
typedef struct SherpaOnnxOfflineModelConfig	SherpaOnnxOfflineModelConfig
	Model configuration shared by offline ASR recognizers.
typedef struct SherpaOnnxOfflineRecognizerConfig	SherpaOnnxOfflineRecognizerConfig
	Configuration for a non-streaming ASR recognizer.
typedef struct SherpaOnnxOfflineRecognizer	SherpaOnnxOfflineRecognizer
	Non-streaming recognizer handle.
typedef struct SherpaOnnxOfflineStream	SherpaOnnxOfflineStream
	Non-streaming decoding state for one utterance.
typedef struct SherpaOnnxOfflineRecognizerResult	SherpaOnnxOfflineRecognizerResult
	Recognition result for a non-streaming ASR stream.
typedef struct SherpaOnnxKeywordResult	SherpaOnnxKeywordResult
	Snapshot of the current keyword spotting result.
typedef struct SherpaOnnxKeywordSpotterConfig	SherpaOnnxKeywordSpotterConfig
	Configuration for keyword spotting.
typedef struct SherpaOnnxKeywordSpotter	SherpaOnnxKeywordSpotter
	Opaque keyword spotter handle.
typedef struct SherpaOnnxSileroVadModelConfig	SherpaOnnxSileroVadModelConfig
	Configuration for a Silero VAD model.
typedef struct SherpaOnnxTenVadModelConfig	SherpaOnnxTenVadModelConfig
	Configuration for a Ten VAD model.
typedef struct SherpaOnnxVadModelConfig	SherpaOnnxVadModelConfig
	Configuration shared by voice activity detectors.
typedef struct SherpaOnnxCircularBuffer	SherpaOnnxCircularBuffer
	Opaque circular-buffer handle used by helper APIs.
typedef struct SherpaOnnxSpeechSegment	SherpaOnnxSpeechSegment
	One detected speech segment returned by the VAD.
typedef struct SherpaOnnxVoiceActivityDetector	SherpaOnnxVoiceActivityDetector
	Opaque voice activity detector handle.
typedef struct SherpaOnnxOfflineTtsVitsModelConfig	SherpaOnnxOfflineTtsVitsModelConfig
	Configuration for a VITS TTS model.
typedef struct SherpaOnnxOfflineTtsMatchaModelConfig	SherpaOnnxOfflineTtsMatchaModelConfig
	Configuration for a Matcha TTS model.
typedef struct SherpaOnnxOfflineTtsKokoroModelConfig	SherpaOnnxOfflineTtsKokoroModelConfig
	Configuration for a Kokoro TTS model.
typedef struct SherpaOnnxOfflineTtsKittenModelConfig	SherpaOnnxOfflineTtsKittenModelConfig
	Configuration for a Kitten TTS model.
typedef struct SherpaOnnxOfflineTtsZipvoiceModelConfig	SherpaOnnxOfflineTtsZipvoiceModelConfig
	Configuration for a ZipVoice TTS model.
typedef struct SherpaOnnxOfflineTtsPocketModelConfig	SherpaOnnxOfflineTtsPocketModelConfig
	Configuration for a Pocket TTS model.
typedef struct SherpaOnnxOfflineTtsSupertonicModelConfig	SherpaOnnxOfflineTtsSupertonicModelConfig
	Configuration for a Supertonic TTS model.
typedef struct SherpaOnnxOfflineTtsModelConfig	SherpaOnnxOfflineTtsModelConfig
	Configuration shared by offline TTS models.
typedef struct SherpaOnnxOfflineTtsConfig	SherpaOnnxOfflineTtsConfig
	Configuration for offline text-to-speech.
typedef struct SherpaOnnxGeneratedAudio	SherpaOnnxGeneratedAudio
	Generated waveform returned by TTS APIs.
typedef int32_t(*	SherpaOnnxGeneratedAudioCallback) (const float *samples, int32_t n)
	Callback invoked during incremental generation.
typedef int32_t(*	SherpaOnnxGeneratedAudioCallbackWithArg) (const float *samples, int32_t n, void *arg)
	Same as SherpaOnnxGeneratedAudioCallback but with an extra user pointer.
typedef int32_t(*	SherpaOnnxGeneratedAudioProgressCallback) (const float *samples, int32_t n, float p)
	Progress callback invoked during incremental generation.
typedef int32_t(*	SherpaOnnxGeneratedAudioProgressCallbackWithArg) (const float *samples, int32_t n, float p, void *arg)
	Same as SherpaOnnxGeneratedAudioProgressCallback but with an extra user pointer.
typedef struct SherpaOnnxOfflineTts	SherpaOnnxOfflineTts
	Opaque offline TTS handle.
typedef struct SherpaOnnxGenerationConfig	SherpaOnnxGenerationConfig
	Generation-time parameters shared by advanced TTS APIs.
typedef struct SherpaOnnxWave	SherpaOnnxWave
	Decoded mono WAVE file content.
typedef struct SherpaOnnxMultiChannelWave	SherpaOnnxMultiChannelWave
	Decoded multi-channel WAVE file content.
typedef struct SherpaOnnxSpokenLanguageIdentificationWhisperConfig	SherpaOnnxSpokenLanguageIdentificationWhisperConfig
	Whisper-based model files for spoken language identification.
typedef struct SherpaOnnxSpokenLanguageIdentificationConfig	SherpaOnnxSpokenLanguageIdentificationConfig
	Configuration for spoken language identification.
typedef struct SherpaOnnxSpokenLanguageIdentification	SherpaOnnxSpokenLanguageIdentification
	Opaque spoken-language identification handle.
typedef struct SherpaOnnxSpokenLanguageIdentificationResult	SherpaOnnxSpokenLanguageIdentificationResult
	Result of spoken-language identification.
typedef struct SherpaOnnxSpeakerEmbeddingExtractorConfig	SherpaOnnxSpeakerEmbeddingExtractorConfig
	Configuration for speaker embedding extraction.
typedef struct SherpaOnnxSpeakerEmbeddingExtractor	SherpaOnnxSpeakerEmbeddingExtractor
	Opaque speaker embedding extractor handle.
typedef struct SherpaOnnxSpeakerEmbeddingManager	SherpaOnnxSpeakerEmbeddingManager
	Opaque speaker embedding manager handle.
typedef struct SherpaOnnxSpeakerEmbeddingManagerSpeakerMatch	SherpaOnnxSpeakerEmbeddingManagerSpeakerMatch
	One speaker match returned by the best-matches API.
typedef struct SherpaOnnxSpeakerEmbeddingManagerBestMatchesResult	SherpaOnnxSpeakerEmbeddingManagerBestMatchesResult
	Collection of best speaker matches.
typedef struct SherpaOnnxOfflineZipformerAudioTaggingModelConfig	SherpaOnnxOfflineZipformerAudioTaggingModelConfig
	Zipformer audio-tagging model configuration.
typedef struct SherpaOnnxAudioTaggingModelConfig	SherpaOnnxAudioTaggingModelConfig
	Audio-tagging model configuration.
typedef struct SherpaOnnxAudioTaggingConfig	SherpaOnnxAudioTaggingConfig
	Configuration for audio tagging.
typedef struct SherpaOnnxAudioEvent	SherpaOnnxAudioEvent
	One audio-tagging prediction.
typedef struct SherpaOnnxAudioTagging	SherpaOnnxAudioTagging
	Opaque audio tagger handle.
typedef struct SherpaOnnxOfflinePunctuationModelConfig	SherpaOnnxOfflinePunctuationModelConfig
	Offline punctuation model configuration.
typedef struct SherpaOnnxOfflinePunctuationConfig	SherpaOnnxOfflinePunctuationConfig
	Configuration for offline punctuation.
typedef struct SherpaOnnxOfflinePunctuation	SherpaOnnxOfflinePunctuation
	Opaque offline punctuation handle.
typedef struct SherpaOnnxOnlinePunctuationModelConfig	SherpaOnnxOnlinePunctuationModelConfig
	Online punctuation model configuration.
typedef struct SherpaOnnxOnlinePunctuationConfig	SherpaOnnxOnlinePunctuationConfig
	Configuration for online punctuation.
typedef struct SherpaOnnxOnlinePunctuation	SherpaOnnxOnlinePunctuation
	Opaque online punctuation handle.
typedef struct SherpaOnnxLinearResampler	SherpaOnnxLinearResampler
	Opaque linear resampler handle.
typedef struct SherpaOnnxResampleOut	SherpaOnnxResampleOut
	Output chunk returned by SherpaOnnxLinearResamplerResample().
typedef struct SherpaOnnxOfflineSpeakerSegmentationPyannoteModelConfig	SherpaOnnxOfflineSpeakerSegmentationPyannoteModelConfig
	Pyannote speaker-segmentation model configuration.
typedef struct SherpaOnnxOfflineSpeakerSegmentationModelConfig	SherpaOnnxOfflineSpeakerSegmentationModelConfig
	Segmentation model configuration for offline speaker diarization.
typedef struct SherpaOnnxFastClusteringConfig	SherpaOnnxFastClusteringConfig
	Fast clustering configuration.
typedef struct SherpaOnnxOfflineSpeakerDiarizationConfig	SherpaOnnxOfflineSpeakerDiarizationConfig
	Configuration for offline speaker diarization.
typedef struct SherpaOnnxOfflineSpeakerDiarization	SherpaOnnxOfflineSpeakerDiarization
	Opaque offline speaker diarization handle.
typedef struct SherpaOnnxOfflineSpeakerDiarizationResult	SherpaOnnxOfflineSpeakerDiarizationResult
	Opaque offline speaker diarization result.
typedef struct SherpaOnnxOfflineSpeakerDiarizationSegment	SherpaOnnxOfflineSpeakerDiarizationSegment
	One diarization segment.
typedef int32_t(*	SherpaOnnxOfflineSpeakerDiarizationProgressCallback) (int32_t num_processed_chunks, int32_t num_total_chunks, void *arg)
	Progress callback for offline speaker diarization.
typedef int32_t(*	SherpaOnnxOfflineSpeakerDiarizationProgressCallbackNoArg) (int32_t num_processed_chunks, int32_t num_total_chunks)
	Same as SherpaOnnxOfflineSpeakerDiarizationProgressCallback but without a user pointer.
typedef struct SherpaOnnxOfflineSpeechDenoiserGtcrnModelConfig	SherpaOnnxOfflineSpeechDenoiserGtcrnModelConfig
	GTCRN offline denoiser model configuration.
typedef struct SherpaOnnxOfflineSpeechDenoiserDpdfNetModelConfig	SherpaOnnxOfflineSpeechDenoiserDpdfNetModelConfig
	DPDFNet offline denoiser model configuration.
typedef struct SherpaOnnxOfflineSpeechDenoiserModelConfig	SherpaOnnxOfflineSpeechDenoiserModelConfig
	Speech denoiser model configuration shared by offline and online APIs.
typedef struct SherpaOnnxOfflineSpeechDenoiserConfig	SherpaOnnxOfflineSpeechDenoiserConfig
	Configuration for offline speech denoising.
typedef struct SherpaOnnxOfflineSpeechDenoiser	SherpaOnnxOfflineSpeechDenoiser
	Opaque offline speech denoiser handle.
typedef struct SherpaOnnxDenoisedAudio	SherpaOnnxDenoisedAudio
	Denoised audio returned by offline or online speech enhancement APIs.
typedef struct SherpaOnnxOnlineSpeechDenoiserConfig	SherpaOnnxOnlineSpeechDenoiserConfig
	Configuration for streaming speech denoising.
typedef struct SherpaOnnxOnlineSpeechDenoiser	SherpaOnnxOnlineSpeechDenoiser
	Opaque online speech denoiser handle.
typedef struct SherpaOnnxOfflineSourceSeparationSpleeterModelConfig	SherpaOnnxOfflineSourceSeparationSpleeterModelConfig
	Spleeter source-separation model configuration.
typedef struct SherpaOnnxOfflineSourceSeparationUvrModelConfig	SherpaOnnxOfflineSourceSeparationUvrModelConfig
	UVR (MDX-Net) source-separation model configuration.
typedef struct SherpaOnnxOfflineSourceSeparationModelConfig	SherpaOnnxOfflineSourceSeparationModelConfig
	Source-separation model configuration.
typedef struct SherpaOnnxOfflineSourceSeparationConfig	SherpaOnnxOfflineSourceSeparationConfig
	Top-level source-separation configuration.
typedef struct SherpaOnnxOfflineSourceSeparation	SherpaOnnxOfflineSourceSeparation
	Opaque source-separation engine handle.
typedef struct SherpaOnnxSourceSeparationStem	SherpaOnnxSourceSeparationStem
	A single stem (one output track) with one or more channels.
typedef struct SherpaOnnxSourceSeparationOutput	SherpaOnnxSourceSeparationOutput
	Output of a source-separation run.

Functions
const char *	SherpaOnnxGetVersionStr ()
	Return the sherpa-onnx version string.
const char *	SherpaOnnxGetGitSha1 ()
	Return the Git SHA1 used to build the library.
const char *	SherpaOnnxGetGitDate ()
	Return the Git build date used to build the library.
int32_t	SherpaOnnxFileExists (const char *filename)
	Check whether a file exists.
const SherpaOnnxOnlineRecognizer *	SherpaOnnxCreateOnlineRecognizer (const SherpaOnnxOnlineRecognizerConfig *config)
	Create a streaming ASR recognizer.
void	SherpaOnnxDestroyOnlineRecognizer (const SherpaOnnxOnlineRecognizer *recognizer)
	Destroy a streaming recognizer.
const SherpaOnnxOnlineStream *	SherpaOnnxCreateOnlineStream (const SherpaOnnxOnlineRecognizer *recognizer)
	Create a streaming ASR state object.
const SherpaOnnxOnlineStream *	SherpaOnnxCreateOnlineStreamWithHotwords (const SherpaOnnxOnlineRecognizer *recognizer, const char *hotwords)
	Create a streaming ASR state object with per-stream hotwords.
void	SherpaOnnxDestroyOnlineStream (const SherpaOnnxOnlineStream *stream)
	Destroy a streaming ASR state object.
void	SherpaOnnxOnlineStreamAcceptWaveform (const SherpaOnnxOnlineStream *stream, int32_t sample_rate, const float *samples, int32_t n)
	Append audio samples to a streaming ASR stream.
int32_t	SherpaOnnxIsOnlineStreamReady (const SherpaOnnxOnlineRecognizer *recognizer, const SherpaOnnxOnlineStream *stream)
	Check whether a streaming ASR stream is ready to decode.
void	SherpaOnnxDecodeOnlineStream (const SherpaOnnxOnlineRecognizer *recognizer, const SherpaOnnxOnlineStream *stream)
	Decode one step of a streaming ASR stream.
void	SherpaOnnxDecodeMultipleOnlineStreams (const SherpaOnnxOnlineRecognizer *recognizer, const SherpaOnnxOnlineStream **streams, int32_t n)
	Decode multiple streaming ASR streams in parallel.
const SherpaOnnxOnlineRecognizerResult *	SherpaOnnxGetOnlineStreamResult (const SherpaOnnxOnlineRecognizer *recognizer, const SherpaOnnxOnlineStream *stream)
	Get the current streaming ASR result for a stream.
void	SherpaOnnxDestroyOnlineRecognizerResult (const SherpaOnnxOnlineRecognizerResult *r)
	Destroy a result returned by SherpaOnnxGetOnlineStreamResult().
const char *	SherpaOnnxGetOnlineStreamResultAsJson (const SherpaOnnxOnlineRecognizer *recognizer, const SherpaOnnxOnlineStream *stream)
	Get the current streaming ASR result as JSON.
void	SherpaOnnxDestroyOnlineStreamResultJson (const char *s)
	Free a JSON string returned by SherpaOnnxGetOnlineStreamResultAsJson().
void	SherpaOnnxOnlineStreamReset (const SherpaOnnxOnlineRecognizer *recognizer, const SherpaOnnxOnlineStream *stream)
	Reset a streaming ASR stream after an endpoint or utterance boundary.
void	SherpaOnnxOnlineStreamInputFinished (const SherpaOnnxOnlineStream *stream)
	Signal end-of-input for a streaming ASR stream.
void	SherpaOnnxOnlineStreamSetOption (const SherpaOnnxOnlineStream *stream, const char *key, const char *value)
	Set a per-stream runtime option.
const char *	SherpaOnnxOnlineStreamGetOption (const SherpaOnnxOnlineStream *stream, const char *key)
	Get a per-stream runtime option.
int32_t	SherpaOnnxOnlineStreamHasOption (const SherpaOnnxOnlineStream *stream, const char *key)
	Check whether a per-stream runtime option exists.
int32_t	SherpaOnnxOnlineStreamIsEndpoint (const SherpaOnnxOnlineRecognizer *recognizer, const SherpaOnnxOnlineStream *stream)
	Check whether endpoint detection has triggered for a stream.
const SherpaOnnxDisplay *	SherpaOnnxCreateDisplay (int32_t max_word_per_line)
	Create a display helper.
void	SherpaOnnxDestroyDisplay (const SherpaOnnxDisplay *display)
	Destroy a display helper.
void	SherpaOnnxPrint (const SherpaOnnxDisplay *display, int32_t idx, const char *s)
	Print one line of text using the display helper.
const SherpaOnnxOfflineRecognizer *	SherpaOnnxCreateOfflineRecognizer (const SherpaOnnxOfflineRecognizerConfig *config)
	Create a non-streaming ASR recognizer.
void	SherpaOnnxOfflineRecognizerSetConfig (const SherpaOnnxOfflineRecognizer *recognizer, const SherpaOnnxOfflineRecognizerConfig *config)
	Update the configuration of an existing offline recognizer.
void	SherpaOnnxDestroyOfflineRecognizer (const SherpaOnnxOfflineRecognizer *recognizer)
	Destroy a non-streaming recognizer.
const SherpaOnnxOfflineStream *	SherpaOnnxCreateOfflineStream (const SherpaOnnxOfflineRecognizer *recognizer)
	Create a non-streaming ASR input stream.
const SherpaOnnxOfflineStream *	SherpaOnnxCreateOfflineStreamWithHotwords (const SherpaOnnxOfflineRecognizer *recognizer, const char *hotwords)
	Create a non-streaming ASR input stream with per-stream hotwords.
void	SherpaOnnxDestroyOfflineStream (const SherpaOnnxOfflineStream *stream)
	Destroy a non-streaming ASR stream.
void	SherpaOnnxAcceptWaveformOffline (const SherpaOnnxOfflineStream *stream, int32_t sample_rate, const float *samples, int32_t n)
	Provide the full utterance to an offline ASR stream.
void	SherpaOnnxOfflineStreamSetOption (const SherpaOnnxOfflineStream *stream, const char *key, const char *value)
	Set a per-stream runtime option for offline ASR.
const char *	SherpaOnnxOfflineStreamGetOption (const SherpaOnnxOfflineStream *stream, const char *key)
	Get a per-stream runtime option for offline ASR.
int32_t	SherpaOnnxOfflineStreamHasOption (const SherpaOnnxOfflineStream *stream, const char *key)
	Check whether a per-stream runtime option exists.
void	SherpaOnnxDecodeOfflineStream (const SherpaOnnxOfflineRecognizer *recognizer, const SherpaOnnxOfflineStream *stream)
	Run offline ASR on one stream.
void	SherpaOnnxDecodeMultipleOfflineStreams (const SherpaOnnxOfflineRecognizer *recognizer, const SherpaOnnxOfflineStream **streams, int32_t n)
	Run offline ASR on multiple streams in parallel.
const SherpaOnnxOfflineRecognizerResult *	SherpaOnnxGetOfflineStreamResult (const SherpaOnnxOfflineStream *stream)
	Get the recognition result for an offline ASR stream.
void	SherpaOnnxDestroyOfflineRecognizerResult (const SherpaOnnxOfflineRecognizerResult *r)
	Destroy a result returned by SherpaOnnxGetOfflineStreamResult().
const char *	SherpaOnnxGetOfflineStreamResultAsJson (const SherpaOnnxOfflineStream *stream)
	Get the offline ASR result as JSON.
void	SherpaOnnxDestroyOfflineStreamResultJson (const char *s)
	Free a JSON string returned by SherpaOnnxGetOfflineStreamResultAsJson().
const SherpaOnnxKeywordSpotter *	SherpaOnnxCreateKeywordSpotter (const SherpaOnnxKeywordSpotterConfig *config)
	Create a keyword spotter.
void	SherpaOnnxDestroyKeywordSpotter (const SherpaOnnxKeywordSpotter *spotter)
	Destroy a keyword spotter.
const SherpaOnnxOnlineStream *	SherpaOnnxCreateKeywordStream (const SherpaOnnxKeywordSpotter *spotter)
	Create a keyword spotting stream using the spotter's built-in keyword list.
const SherpaOnnxOnlineStream *	SherpaOnnxCreateKeywordStreamWithKeywords (const SherpaOnnxKeywordSpotter *spotter, const char *keywords)
	Create a keyword spotting stream with extra or replacement keywords.
int32_t	SherpaOnnxIsKeywordStreamReady (const SherpaOnnxKeywordSpotter *spotter, const SherpaOnnxOnlineStream *stream)
	Check whether a keyword stream has enough audio for decoding.
void	SherpaOnnxDecodeKeywordStream (const SherpaOnnxKeywordSpotter *spotter, const SherpaOnnxOnlineStream *stream)
	Decode one ready keyword stream.
void	SherpaOnnxResetKeywordStream (const SherpaOnnxKeywordSpotter *spotter, const SherpaOnnxOnlineStream *stream)
	Reset a keyword stream after a keyword is detected.
void	SherpaOnnxDecodeMultipleKeywordStreams (const SherpaOnnxKeywordSpotter *spotter, const SherpaOnnxOnlineStream **streams, int32_t n)
	Decode multiple ready keyword streams in parallel.
const SherpaOnnxKeywordResult *	SherpaOnnxGetKeywordResult (const SherpaOnnxKeywordSpotter *spotter, const SherpaOnnxOnlineStream *stream)
	Get the current keyword spotting result for a stream.
void	SherpaOnnxDestroyKeywordResult (const SherpaOnnxKeywordResult *r)
	Destroy a keyword result snapshot.
const char *	SherpaOnnxGetKeywordResultAsJson (const SherpaOnnxKeywordSpotter *spotter, const SherpaOnnxOnlineStream *stream)
	Get the current keyword spotting result as JSON.
void	SherpaOnnxFreeKeywordResultJson (const char *s)
	Free a JSON string returned by SherpaOnnxGetKeywordResultAsJson().
const SherpaOnnxCircularBuffer *	SherpaOnnxCreateCircularBuffer (int32_t capacity)
	Create a floating-point circular buffer.
void	SherpaOnnxDestroyCircularBuffer (const SherpaOnnxCircularBuffer *buffer)
	Destroy a circular buffer.
void	SherpaOnnxCircularBufferPush (const SherpaOnnxCircularBuffer *buffer, const float *p, int32_t n)
	Append samples to a circular buffer.
const float *	SherpaOnnxCircularBufferGet (const SherpaOnnxCircularBuffer *buffer, int32_t start_index, int32_t n)
	Copy out a slice of samples from a circular buffer.
void	SherpaOnnxCircularBufferFree (const float *p)
	Free an array returned by SherpaOnnxCircularBufferGet().
void	SherpaOnnxCircularBufferPop (const SherpaOnnxCircularBuffer *buffer, int32_t n)
	Drop samples from the front of a circular buffer.
int32_t	SherpaOnnxCircularBufferSize (const SherpaOnnxCircularBuffer *buffer)
	Return the number of currently stored samples.
int32_t	SherpaOnnxCircularBufferHead (const SherpaOnnxCircularBuffer *buffer)
	Return the current head index of the buffer timeline.
void	SherpaOnnxCircularBufferReset (const SherpaOnnxCircularBuffer *buffer)
	Clear a circular buffer and reset its head index.
const SherpaOnnxVoiceActivityDetector *	SherpaOnnxCreateVoiceActivityDetector (const SherpaOnnxVadModelConfig *config, float buffer_size_in_seconds)
	Create a voice activity detector.
void	SherpaOnnxDestroyVoiceActivityDetector (const SherpaOnnxVoiceActivityDetector *p)
	Destroy a voice activity detector.
void	SherpaOnnxVoiceActivityDetectorAcceptWaveform (const SherpaOnnxVoiceActivityDetector *p, const float *samples, int32_t n)
	Feed audio samples to the VAD.
int32_t	SherpaOnnxVoiceActivityDetectorEmpty (const SherpaOnnxVoiceActivityDetector *p)
	Check whether the detector currently has any completed speech segment.
int32_t	SherpaOnnxVoiceActivityDetectorDetected (const SherpaOnnxVoiceActivityDetector *p)
	Check whether the detector is currently inside speech.
void	SherpaOnnxVoiceActivityDetectorPop (const SherpaOnnxVoiceActivityDetector *p)
	Remove the front speech segment from the detector queue.
void	SherpaOnnxVoiceActivityDetectorClear (const SherpaOnnxVoiceActivityDetector *p)
	Remove all queued speech segments.
const SherpaOnnxSpeechSegment *	SherpaOnnxVoiceActivityDetectorFront (const SherpaOnnxVoiceActivityDetector *p)
	Get the first queued speech segment.
void	SherpaOnnxDestroySpeechSegment (const SherpaOnnxSpeechSegment *p)
	Destroy a speech segment returned by SherpaOnnxVoiceActivityDetectorFront().
void	SherpaOnnxVoiceActivityDetectorReset (const SherpaOnnxVoiceActivityDetector *p)
	Reset a voice activity detector so it can process a new stream.
void	SherpaOnnxVoiceActivityDetectorFlush (const SherpaOnnxVoiceActivityDetector *p)
	Flush buffered tail samples and force final segmentation.
const SherpaOnnxOfflineTts *	SherpaOnnxCreateOfflineTts (const SherpaOnnxOfflineTtsConfig *config)
	Create an offline TTS engine.
void	SherpaOnnxDestroyOfflineTts (const SherpaOnnxOfflineTts *tts)
	Destroy an offline TTS engine.
int32_t	SherpaOnnxOfflineTtsSampleRate (const SherpaOnnxOfflineTts *tts)
	Return the output sample rate of a TTS engine.
int32_t	SherpaOnnxOfflineTtsNumSpeakers (const SherpaOnnxOfflineTts *tts)
	Return the number of available speaker IDs.
const SherpaOnnxGeneratedAudio *	SherpaOnnxOfflineTtsGenerate (const SherpaOnnxOfflineTts *tts, const char *text, int32_t sid, float speed)
	Generate speech from text using the simple sid/speed interface.
const SherpaOnnxGeneratedAudio *	SherpaOnnxOfflineTtsGenerateWithCallback (const SherpaOnnxOfflineTts *tts, const char *text, int32_t sid, float speed, SherpaOnnxGeneratedAudioCallback callback)
	Generate speech and receive incremental audio chunks through a callback.
const SherpaOnnxGeneratedAudio *	SherpaOnnxOfflineTtsGenerateWithProgressCallback (const SherpaOnnxOfflineTts *tts, const char *text, int32_t sid, float speed, SherpaOnnxGeneratedAudioProgressCallback callback)
	Generate speech with a progress callback.
const SherpaOnnxGeneratedAudio *	SherpaOnnxOfflineTtsGenerateWithProgressCallbackWithArg (const SherpaOnnxOfflineTts *tts, const char *text, int32_t sid, float speed, SherpaOnnxGeneratedAudioProgressCallbackWithArg callback, void *arg)
	Generate speech with a progress callback that receives a user pointer.
const SherpaOnnxGeneratedAudio *	SherpaOnnxOfflineTtsGenerateWithCallbackWithArg (const SherpaOnnxOfflineTts *tts, const char *text, int32_t sid, float speed, SherpaOnnxGeneratedAudioCallbackWithArg callback, void *arg)
	Same as SherpaOnnxOfflineTtsGenerateWithCallback() but with a user pointer.
const SherpaOnnxGeneratedAudio *	SherpaOnnxOfflineTtsGenerateWithZipvoice (const SherpaOnnxOfflineTts *tts, const char *text, const char *prompt_text, const float *prompt_samples, int32_t n_prompt, int32_t prompt_sr, float speed, int32_t num_steps)
	Deprecated ZipVoice-specific generation API.
const SherpaOnnxGeneratedAudio *	SherpaOnnxOfflineTtsGenerateWithConfig (const SherpaOnnxOfflineTts *tts, const char *text, const SherpaOnnxGenerationConfig *config, SherpaOnnxGeneratedAudioProgressCallbackWithArg callback, void *arg)
	Generate speech using the advanced configuration interface.
void	SherpaOnnxDestroyOfflineTtsGeneratedAudio (const SherpaOnnxGeneratedAudio *p)
	Destroy audio returned by a TTS generation API.
int32_t	SherpaOnnxWriteWave (const float *samples, int32_t n, int32_t sample_rate, const char *filename)
	Write floating-point PCM to a mono 16-bit WAVE file.
int64_t	SherpaOnnxWaveFileSize (int32_t n_samples)
	Return the number of bytes needed for a mono 16-bit WAVE file.
void	SherpaOnnxWriteWaveToBuffer (const float *samples, int32_t n, int32_t sample_rate, char *buffer)
	Write a mono 16-bit WAVE file to a caller-provided buffer.
int32_t	SherpaOnnxWriteWaveMultiChannel (const float *const *samples, int32_t n, int32_t sample_rate, int32_t num_channels, const char *filename)
	Write multi-channel audio to a WAVE file (16-bit PCM).
const SherpaOnnxWave *	SherpaOnnxReadWave (const char *filename)
	Read a mono 16-bit PCM WAVE file.
const SherpaOnnxWave *	SherpaOnnxReadWaveFromBinaryData (const char *data, int32_t n)
	Read a mono 16-bit PCM WAVE file from binary memory.
void	SherpaOnnxFreeWave (const SherpaOnnxWave *wave)
	Destroy a wave object returned by SherpaOnnxReadWave() or SherpaOnnxReadWaveFromBinaryData().
const SherpaOnnxMultiChannelWave *	SherpaOnnxReadWaveMultiChannel (const char *filename)
	Read a multi-channel 16-bit PCM WAVE file.
void	SherpaOnnxFreeMultiChannelWave (const SherpaOnnxMultiChannelWave *wave)
	Destroy a multi-channel wave object.
const SherpaOnnxSpokenLanguageIdentification *	SherpaOnnxCreateSpokenLanguageIdentification (const SherpaOnnxSpokenLanguageIdentificationConfig *config)
	Create a spoken-language identifier.
void	SherpaOnnxDestroySpokenLanguageIdentification (const SherpaOnnxSpokenLanguageIdentification *slid)
	Destroy a spoken-language identifier.
SherpaOnnxOfflineStream *	SherpaOnnxSpokenLanguageIdentificationCreateOfflineStream (const SherpaOnnxSpokenLanguageIdentification *slid)
	Create an offline stream for spoken-language identification.
const SherpaOnnxSpokenLanguageIdentificationResult *	SherpaOnnxSpokenLanguageIdentificationCompute (const SherpaOnnxSpokenLanguageIdentification *slid, const SherpaOnnxOfflineStream *s)
	Run spoken-language identification on an offline stream.
void	SherpaOnnxDestroySpokenLanguageIdentificationResult (const SherpaOnnxSpokenLanguageIdentificationResult *r)
	Destroy a spoken-language identification result.
const SherpaOnnxSpeakerEmbeddingExtractor *	SherpaOnnxCreateSpeakerEmbeddingExtractor (const SherpaOnnxSpeakerEmbeddingExtractorConfig *config)
	Create a speaker embedding extractor.
void	SherpaOnnxDestroySpeakerEmbeddingExtractor (const SherpaOnnxSpeakerEmbeddingExtractor *p)
	Destroy a speaker embedding extractor.
int32_t	SherpaOnnxSpeakerEmbeddingExtractorDim (const SherpaOnnxSpeakerEmbeddingExtractor *p)
	Return the embedding dimension produced by the extractor.
const SherpaOnnxOnlineStream *	SherpaOnnxSpeakerEmbeddingExtractorCreateStream (const SherpaOnnxSpeakerEmbeddingExtractor *p)
	Create a streaming feature buffer for embedding extraction.
int32_t	SherpaOnnxSpeakerEmbeddingExtractorIsReady (const SherpaOnnxSpeakerEmbeddingExtractor *p, const SherpaOnnxOnlineStream *s)
	Check whether enough audio has been provided to compute an embedding.
const float *	SherpaOnnxSpeakerEmbeddingExtractorComputeEmbedding (const SherpaOnnxSpeakerEmbeddingExtractor *p, const SherpaOnnxOnlineStream *s)
	Compute the embedding for a stream.
void	SherpaOnnxSpeakerEmbeddingExtractorDestroyEmbedding (const float *v)
	Destroy an embedding vector returned by SherpaOnnxSpeakerEmbeddingExtractorComputeEmbedding().
const SherpaOnnxSpeakerEmbeddingManager *	SherpaOnnxCreateSpeakerEmbeddingManager (int32_t dim)
	Create a speaker embedding manager.
void	SherpaOnnxDestroySpeakerEmbeddingManager (const SherpaOnnxSpeakerEmbeddingManager *p)
	Destroy a speaker embedding manager.
int32_t	SherpaOnnxSpeakerEmbeddingManagerAdd (const SherpaOnnxSpeakerEmbeddingManager *p, const char *name, const float *v)
	Add one enrollment embedding for a speaker.
int32_t	SherpaOnnxSpeakerEmbeddingManagerAddList (const SherpaOnnxSpeakerEmbeddingManager *p, const char *name, const float **v)
	Add multiple enrollment embeddings for one speaker.
int32_t	SherpaOnnxSpeakerEmbeddingManagerAddListFlattened (const SherpaOnnxSpeakerEmbeddingManager *p, const char *name, const float *v, int32_t n)
	Add multiple enrollment embeddings packed in one flat array.
int32_t	SherpaOnnxSpeakerEmbeddingManagerRemove (const SherpaOnnxSpeakerEmbeddingManager *p, const char *name)
	Remove a speaker from the manager.
const char *	SherpaOnnxSpeakerEmbeddingManagerSearch (const SherpaOnnxSpeakerEmbeddingManager *p, const float *v, float threshold)
	Search for the best matching enrolled speaker.
void	SherpaOnnxSpeakerEmbeddingManagerFreeSearch (const char *name)
	Free a string returned by SherpaOnnxSpeakerEmbeddingManagerSearch().
const SherpaOnnxSpeakerEmbeddingManagerBestMatchesResult *	SherpaOnnxSpeakerEmbeddingManagerGetBestMatches (const SherpaOnnxSpeakerEmbeddingManager *p, const float *v, float threshold, int32_t n)
	Return up to n best matches above a similarity threshold.
void	SherpaOnnxSpeakerEmbeddingManagerFreeBestMatches (const SherpaOnnxSpeakerEmbeddingManagerBestMatchesResult *r)
	Destroy a best-matches result.
int32_t	SherpaOnnxSpeakerEmbeddingManagerVerify (const SherpaOnnxSpeakerEmbeddingManager *p, const char *name, const float *v, float threshold)
	Verify whether a query embedding matches a named speaker.
int32_t	SherpaOnnxSpeakerEmbeddingManagerContains (const SherpaOnnxSpeakerEmbeddingManager *p, const char *name)
	Check whether a speaker is enrolled.
int32_t	SherpaOnnxSpeakerEmbeddingManagerNumSpeakers (const SherpaOnnxSpeakerEmbeddingManager *p)
	Return the number of enrolled speakers.
const char *const *	SherpaOnnxSpeakerEmbeddingManagerGetAllSpeakers (const SherpaOnnxSpeakerEmbeddingManager *p)
	Return all enrolled speaker names.
void	SherpaOnnxSpeakerEmbeddingManagerFreeAllSpeakers (const char *const *names)
	Free an array returned by SherpaOnnxSpeakerEmbeddingManagerGetAllSpeakers().
const SherpaOnnxAudioTagging *	SherpaOnnxCreateAudioTagging (const SherpaOnnxAudioTaggingConfig *config)
	Create an audio tagger.
void	SherpaOnnxDestroyAudioTagging (const SherpaOnnxAudioTagging *tagger)
	Destroy an audio tagger.
const SherpaOnnxOfflineStream *	SherpaOnnxAudioTaggingCreateOfflineStream (const SherpaOnnxAudioTagging *tagger)
	Create an offline stream for audio tagging.
const SherpaOnnxAudioEvent *const *	SherpaOnnxAudioTaggingCompute (const SherpaOnnxAudioTagging *tagger, const SherpaOnnxOfflineStream *s, int32_t top_k)
	Run audio tagging on an offline stream.
void	SherpaOnnxAudioTaggingFreeResults (const SherpaOnnxAudioEvent *const *p)
	Destroy results returned by SherpaOnnxAudioTaggingCompute().
const SherpaOnnxOfflinePunctuation *	SherpaOnnxCreateOfflinePunctuation (const SherpaOnnxOfflinePunctuationConfig *config)
	Create an offline punctuation processor.
void	SherpaOnnxDestroyOfflinePunctuation (const SherpaOnnxOfflinePunctuation *punct)
	Destroy an offline punctuation processor.
const char *	SherpaOfflinePunctuationAddPunct (const SherpaOnnxOfflinePunctuation *punct, const char *text)
	Add punctuation to a complete input text.
void	SherpaOfflinePunctuationFreeText (const char *text)
	Free a string returned by SherpaOfflinePunctuationAddPunct().
const SherpaOnnxOnlinePunctuation *	SherpaOnnxCreateOnlinePunctuation (const SherpaOnnxOnlinePunctuationConfig *config)
	Create an online punctuation processor.
void	SherpaOnnxDestroyOnlinePunctuation (const SherpaOnnxOnlinePunctuation *punctuation)
	Destroy an online punctuation processor.
const char *	SherpaOnnxOnlinePunctuationAddPunct (const SherpaOnnxOnlinePunctuation *punctuation, const char *text)
	Add punctuation to one text chunk using the online punctuation model.
void	SherpaOnnxOnlinePunctuationFreeText (const char *text)
	Free a string returned by SherpaOnnxOnlinePunctuationAddPunct().
const SherpaOnnxLinearResampler *	SherpaOnnxCreateLinearResampler (int32_t samp_rate_in_hz, int32_t samp_rate_out_hz, float filter_cutoff_hz, int32_t num_zeros)
	Create a linear resampler.
void	SherpaOnnxDestroyLinearResampler (const SherpaOnnxLinearResampler *p)
	Destroy a linear resampler.
void	SherpaOnnxLinearResamplerReset (const SherpaOnnxLinearResampler *p)
	Reset a linear resampler to its initial state.
const SherpaOnnxResampleOut *	SherpaOnnxLinearResamplerResample (const SherpaOnnxLinearResampler *p, const float *input, int32_t input_dim, int32_t flush)
	Resample one chunk of input audio.
void	SherpaOnnxLinearResamplerResampleFree (const SherpaOnnxResampleOut *p)
	Destroy a resampler output chunk.
int32_t	SherpaOnnxLinearResamplerResampleGetInputSampleRate (const SherpaOnnxLinearResampler *p)
	Return the resampler input sample rate.
int32_t	SherpaOnnxLinearResamplerResampleGetOutputSampleRate (const SherpaOnnxLinearResampler *p)
	Return the resampler output sample rate.
const SherpaOnnxOfflineSpeakerDiarization *	SherpaOnnxCreateOfflineSpeakerDiarization (const SherpaOnnxOfflineSpeakerDiarizationConfig *config)
	Create an offline speaker diarization pipeline.
void	SherpaOnnxDestroyOfflineSpeakerDiarization (const SherpaOnnxOfflineSpeakerDiarization *sd)
	Destroy an offline speaker diarizer.
int32_t	SherpaOnnxOfflineSpeakerDiarizationGetSampleRate (const SherpaOnnxOfflineSpeakerDiarization *sd)
	Return the expected input sample rate.
void	SherpaOnnxOfflineSpeakerDiarizationSetConfig (const SherpaOnnxOfflineSpeakerDiarization *sd, const SherpaOnnxOfflineSpeakerDiarizationConfig *config)
	Update clustering-related settings of an existing diarizer.
int32_t	SherpaOnnxOfflineSpeakerDiarizationResultGetNumSpeakers (const SherpaOnnxOfflineSpeakerDiarizationResult *r)
	Return the number of speakers in a diarization result.
int32_t	SherpaOnnxOfflineSpeakerDiarizationResultGetNumSegments (const SherpaOnnxOfflineSpeakerDiarizationResult *r)
	Return the number of diarization segments.
const SherpaOnnxOfflineSpeakerDiarizationSegment *	SherpaOnnxOfflineSpeakerDiarizationResultSortByStartTime (const SherpaOnnxOfflineSpeakerDiarizationResult *r)
	Return segments sorted by start time.
void	SherpaOnnxOfflineSpeakerDiarizationDestroySegment (const SherpaOnnxOfflineSpeakerDiarizationSegment *s)
	Destroy a segment array returned by SherpaOnnxOfflineSpeakerDiarizationResultSortByStartTime().
const SherpaOnnxOfflineSpeakerDiarizationResult *	SherpaOnnxOfflineSpeakerDiarizationProcess (const SherpaOnnxOfflineSpeakerDiarization *sd, const float *samples, int32_t n)
	Run offline speaker diarization.
const SherpaOnnxOfflineSpeakerDiarizationResult *	SherpaOnnxOfflineSpeakerDiarizationProcessWithCallback (const SherpaOnnxOfflineSpeakerDiarization *sd, const float *samples, int32_t n, SherpaOnnxOfflineSpeakerDiarizationProgressCallback callback, void *arg)
	Run offline speaker diarization with a progress callback.
const SherpaOnnxOfflineSpeakerDiarizationResult *	SherpaOnnxOfflineSpeakerDiarizationProcessWithCallbackNoArg (const SherpaOnnxOfflineSpeakerDiarization *sd, const float *samples, int32_t n, SherpaOnnxOfflineSpeakerDiarizationProgressCallbackNoArg callback)
	Run offline speaker diarization with a progress callback that has no user pointer.
void	SherpaOnnxOfflineSpeakerDiarizationDestroyResult (const SherpaOnnxOfflineSpeakerDiarizationResult *r)
	Destroy a diarization result.
const SherpaOnnxOfflineSpeechDenoiser *	SherpaOnnxCreateOfflineSpeechDenoiser (const SherpaOnnxOfflineSpeechDenoiserConfig *config)
	Create an offline speech denoiser.
void	SherpaOnnxDestroyOfflineSpeechDenoiser (const SherpaOnnxOfflineSpeechDenoiser *sd)
	Destroy an offline speech denoiser.
int32_t	SherpaOnnxOfflineSpeechDenoiserGetSampleRate (const SherpaOnnxOfflineSpeechDenoiser *sd)
	Return the expected sample rate for the denoiser.
const SherpaOnnxDenoisedAudio *	SherpaOnnxOfflineSpeechDenoiserRun (const SherpaOnnxOfflineSpeechDenoiser *sd, const float *samples, int32_t n, int32_t sample_rate)
	Run offline speech denoising on a complete waveform.
void	SherpaOnnxDestroyDenoisedAudio (const SherpaOnnxDenoisedAudio *p)
	Destroy denoised audio returned by a speech enhancement API.
const SherpaOnnxOnlineSpeechDenoiser *	SherpaOnnxCreateOnlineSpeechDenoiser (const SherpaOnnxOnlineSpeechDenoiserConfig *config)
	Create an online speech denoiser.
void	SherpaOnnxDestroyOnlineSpeechDenoiser (const SherpaOnnxOnlineSpeechDenoiser *sd)
	Destroy an online speech denoiser.
int32_t	SherpaOnnxOnlineSpeechDenoiserGetSampleRate (const SherpaOnnxOnlineSpeechDenoiser *sd)
	Return the expected input sample rate for the online denoiser.
int32_t	SherpaOnnxOnlineSpeechDenoiserGetFrameShiftInSamples (const SherpaOnnxOnlineSpeechDenoiser *sd)
	Return the recommended chunk size in samples for streaming input.
const SherpaOnnxDenoisedAudio *	SherpaOnnxOnlineSpeechDenoiserRun (const SherpaOnnxOnlineSpeechDenoiser *sd, const float *samples, int32_t n, int32_t sample_rate)
	Process one chunk of streaming audio.
const SherpaOnnxDenoisedAudio *	SherpaOnnxOnlineSpeechDenoiserFlush (const SherpaOnnxOnlineSpeechDenoiser *sd)
	Flush buffered samples and reset the online denoiser.
void	SherpaOnnxOnlineSpeechDenoiserReset (const SherpaOnnxOnlineSpeechDenoiser *sd)
	Reset an online denoiser so it can process a new stream.
const SherpaOnnxOfflineSourceSeparation *	SherpaOnnxCreateOfflineSourceSeparation (const SherpaOnnxOfflineSourceSeparationConfig *config)
	Create a source-separation engine.
void	SherpaOnnxDestroyOfflineSourceSeparation (const SherpaOnnxOfflineSourceSeparation *ss)
	Destroy a source-separation engine.
int32_t	SherpaOnnxOfflineSourceSeparationGetOutputSampleRate (const SherpaOnnxOfflineSourceSeparation *ss)
	Return the output sample rate of the source-separation engine.
int32_t	SherpaOnnxOfflineSourceSeparationGetNumberOfStems (const SherpaOnnxOfflineSourceSeparation *ss)
	Return the number of stems produced by the engine.
const SherpaOnnxSourceSeparationOutput *	SherpaOnnxOfflineSourceSeparationProcess (const SherpaOnnxOfflineSourceSeparation *ss, const float *const *samples, int32_t num_channels, int32_t num_samples, int32_t sample_rate)
	Run source separation on multi-channel audio.
void	SherpaOnnxDestroySourceSeparationOutput (const SherpaOnnxSourceSeparationOutput *p)
	Destroy the output of a source-separation run.

Detailed Description

This header exposes the main sherpa-onnx inference features through a stable C interface. It is intended for native C/C++ applications and for language bindings that need a C ABI.

The file is organized by feature family. The major API groups are:

• Utility helpers: version/build information, file checks, WAVE I/O, and a display helper for incremental text output
• Streaming ASR: online recognizers, online streams, endpointing, and per-stream runtime options
• Non-streaming ASR: offline recognizers, offline streams, batch decode, and result retrieval
• Keyword spotting: streaming keyword detection, custom keyword streams, and keyword result snapshots
• Voice activity detection: Silero/Ten VAD models, speech segment buffers, and detector state management
• Text-to-speech: offline TTS model families, generation configuration, and generated audio helpers
• Spoken language identification
• Speaker embedding extraction and speaker enrollment/search/verification
• Audio tagging
• Offline and online punctuation restoration
• Linear resampling
• Offline speaker diarization
• Offline and online speech enhancement / denoising
• HarmonyOS-specific constructor variants

Common ownership rules:

• Opaque handles created by SherpaOnnxCreate*() functions are generally destroyed with a matching SherpaOnnxDestroy*() function
• Snapshot/result objects returned by query functions usually need explicit destruction as documented on each API
• Strings or arrays returned by helper/query functions are either:
  • statically owned by the library and must not be freed, or
  • heap-allocated for the caller and must be released with the matching Free/Destroy API

General usage pattern:

1. Zero-initialize a config struct with memset(&config, 0, sizeof(config))
2. Fill in the required model paths and runtime options
3. Create the corresponding engine with SherpaOnnxCreate*()
4. Create a stream if the feature uses one
5. Feed audio or text, run the compute/decode API, and retrieve results
6. Release every returned object with the documented matching API

The examples in c-api-examples/ show complete end-to-end usage. Useful starting points include:

• decode-file-c-api.c for ASR
• kws-c-api.c for keyword spotting
• vad-whisper-c-api.c for VAD
• offline-tts-c-api.c and kokoro-tts-en-c-api.c for TTS
• speaker-identification-c-api.c for speaker embedding and verification
• audio-tagging-c-api.c for audio tagging
• add-punctuation-c-api.c and add-punctuation-online-c-api.c for punctuation
• offline-sepaker-diarization-c-api.c for diarization
• speech-enhancement-gtcrn-c-api.c and online-speech-enhancement-gtcrn-c-api.c for speech enhancement

Definition in file c-api.h.

Macro Definition Documentation

SHERPA_ONNX_API

#define SHERPA_ONNX_API   SHERPA_ONNX_IMPORT

Definition at line 106 of file c-api.h.

SHERPA_ONNX_EXPORT

#define SHERPA_ONNX_EXPORT   __attribute__((visibility("default")))

Definition at line 98 of file c-api.h.

SHERPA_ONNX_IMPORT

#define SHERPA_ONNX_IMPORT   SHERPA_ONNX_EXPORT

Definition at line 100 of file c-api.h.

Typedef Documentation

SherpaOnnxAudioEvent

typedef struct SherpaOnnxAudioEvent SherpaOnnxAudioEvent

SherpaOnnxAudioTagging

typedef struct SherpaOnnxAudioTagging SherpaOnnxAudioTagging

Definition at line 3438 of file c-api.h.

SherpaOnnxAudioTaggingConfig

typedef struct SherpaOnnxAudioTaggingConfig SherpaOnnxAudioTaggingConfig

SherpaOnnxAudioTaggingConfig config;
memset(&config, 0, sizeof(config));
config.model.zipformer.model =
    "./sherpa-onnx-zipformer-audio-tagging-2024-04-09/model.int8.onnx";
config.model.num_threads = 1;
config.model.provider = "cpu";
config.labels =
    "./sherpa-onnx-zipformer-audio-tagging-2024-04-09/class_labels_indices.csv";
config.top_k = 5;

SherpaOnnxAudioTaggingModelConfig

typedef struct SherpaOnnxAudioTaggingModelConfig SherpaOnnxAudioTaggingModelConfig

Configure exactly one model family. If multiple model families are provided, one of them will be used and the choice is implementation-defined.

Example using sherpa-onnx-zipformer-audio-tagging-2024-04-09:

SherpaOnnxAudioTaggingModelConfig model;
memset(&model, 0, sizeof(model));
model.zipformer.model =
    "./sherpa-onnx-zipformer-audio-tagging-2024-04-09/model.int8.onnx";
model.num_threads = 1;
model.provider = "cpu";

SherpaOnnxCircularBuffer

typedef struct SherpaOnnxCircularBuffer SherpaOnnxCircularBuffer

Definition at line 1936 of file c-api.h.

SherpaOnnxDenoisedAudio

typedef struct SherpaOnnxDenoisedAudio SherpaOnnxDenoisedAudio

Free this object with SherpaOnnxDestroyDenoisedAudio().

SherpaOnnxDisplay

typedef struct SherpaOnnxDisplay SherpaOnnxDisplay

This utility is mainly used by example programs on Linux and macOS.

Definition at line 778 of file c-api.h.

SherpaOnnxFastClusteringConfig

typedef struct SherpaOnnxFastClusteringConfig SherpaOnnxFastClusteringConfig

If num_clusters is greater than 0, threshold is ignored. When the number of speakers is known in advance, setting num_clusters is strongly recommended.

SherpaOnnxFeatureConfig

typedef struct SherpaOnnxFeatureConfig SherpaOnnxFeatureConfig

The bundled ASR models typically expect 16 kHz mono audio and 80-bin features.

SherpaOnnxGeneratedAudio

typedef struct SherpaOnnxGeneratedAudio SherpaOnnxGeneratedAudio

The returned structure owns samples. Free the whole object with SherpaOnnxDestroyOfflineTtsGeneratedAudio().

SherpaOnnxGeneratedAudioCallback

typedef int32_t(* SherpaOnnxGeneratedAudioCallback) (const float *samples, int32_t n)

Return 1 to continue generation. Return 0 to stop early.

The samples pointer is only valid during the callback. Copy the samples if you need to keep them after the callback returns.

Definition at line 2443 of file c-api.h.

SherpaOnnxGeneratedAudioCallbackWithArg

typedef int32_t(* SherpaOnnxGeneratedAudioCallbackWithArg) (const float *samples, int32_t n, void *arg)

Definition at line 2450 of file c-api.h.

SherpaOnnxGeneratedAudioProgressCallback

typedef int32_t(* SherpaOnnxGeneratedAudioProgressCallback) (const float *samples, int32_t n, float p)

Parameters

samples	Newly generated samples valid only during the callback.
n	Number of samples in samples.
p	Progress in the range [0, 1].

Returns

Return 1 to continue generation. Return 0 to stop early.

Definition at line 2462 of file c-api.h.

SherpaOnnxGeneratedAudioProgressCallbackWithArg

typedef int32_t(* SherpaOnnxGeneratedAudioProgressCallbackWithArg) (const float *samples, int32_t n, float p, void *arg)

Definition at line 2469 of file c-api.h.

SherpaOnnxGenerationConfig

typedef struct SherpaOnnxGenerationConfig SherpaOnnxGenerationConfig

This struct supports both simple multi-speaker synthesis and more advanced zero-shot or reference-conditioned models.

Example for Pocket TTS:

SherpaOnnxGenerationConfig cfg;
memset(&cfg, 0, sizeof(cfg));
cfg.speed = 1.0f;
cfg.reference_audio = wave->samples;
cfg.reference_audio_len = wave->num_samples;
cfg.reference_sample_rate = wave->sample_rate;
cfg.extra = "{\"max_reference_audio_len\": 10.0, \"seed\": 42}";

SherpaOnnxHomophoneReplacerConfig

typedef struct SherpaOnnxHomophoneReplacerConfig SherpaOnnxHomophoneReplacerConfig

SherpaOnnxKeywordResult

typedef struct SherpaOnnxKeywordResult SherpaOnnxKeywordResult

Free this object with SherpaOnnxDestroyKeywordResult().

SherpaOnnxKeywordSpotter

typedef struct SherpaOnnxKeywordSpotter SherpaOnnxKeywordSpotter

Definition at line 1685 of file c-api.h.

SherpaOnnxKeywordSpotterConfig

typedef struct SherpaOnnxKeywordSpotterConfig SherpaOnnxKeywordSpotterConfig

The acoustic model is configured through model_config. In practice this is usually a streaming transducer model.

Keyword definitions can be provided either through keywords_file or through keywords_buf/keywords_buf_size. If both are set, the buffer is used.

Example using sherpa-onnx-kws-zipformer-wenetspeech-3.3M-2024-01-01-mobile:

SherpaOnnxKeywordSpotterConfig config;
memset(&config, 0, sizeof(config));
 
config.model_config.transducer.encoder =
    "./sherpa-onnx-kws-zipformer-wenetspeech-3.3M-2024-01-01-mobile/"
    "encoder-epoch-12-avg-2-chunk-16-left-64.int8.onnx";
config.model_config.transducer.decoder =
    "./sherpa-onnx-kws-zipformer-wenetspeech-3.3M-2024-01-01-mobile/"
    "decoder-epoch-12-avg-2-chunk-16-left-64.onnx";
config.model_config.transducer.joiner =
    "./sherpa-onnx-kws-zipformer-wenetspeech-3.3M-2024-01-01-mobile/"
    "joiner-epoch-12-avg-2-chunk-16-left-64.int8.onnx";
config.model_config.tokens =
    "./sherpa-onnx-kws-zipformer-wenetspeech-3.3M-2024-01-01-mobile/"
    "tokens.txt";
config.model_config.provider = "cpu";
config.model_config.num_threads = 1;
 
config.keywords_file =
    "./sherpa-onnx-kws-zipformer-wenetspeech-3.3M-2024-01-01-mobile/"
    "test_wavs/test_keywords.txt";
config.max_active_paths = 4;
config.keywords_score = 3.0f;
config.keywords_threshold = 0.1f;

SherpaOnnxLinearResampler

typedef struct SherpaOnnxLinearResampler SherpaOnnxLinearResampler

Definition at line 3662 of file c-api.h.

SherpaOnnxMultiChannelWave

typedef struct SherpaOnnxMultiChannelWave SherpaOnnxMultiChannelWave

Free this object with SherpaOnnxFreeMultiChannelWave().

SherpaOnnxOfflineCanaryModelConfig

typedef struct SherpaOnnxOfflineCanaryModelConfig SherpaOnnxOfflineCanaryModelConfig

SherpaOnnxOfflineCohereTranscribeModelConfig

typedef struct SherpaOnnxOfflineCohereTranscribeModelConfig SherpaOnnxOfflineCohereTranscribeModelConfig

SherpaOnnxOfflineDolphinModelConfig

typedef struct SherpaOnnxOfflineDolphinModelConfig SherpaOnnxOfflineDolphinModelConfig

SherpaOnnxOfflineFireRedAsrCtcModelConfig

typedef struct SherpaOnnxOfflineFireRedAsrCtcModelConfig SherpaOnnxOfflineFireRedAsrCtcModelConfig

SherpaOnnxOfflineFireRedAsrModelConfig

typedef struct SherpaOnnxOfflineFireRedAsrModelConfig SherpaOnnxOfflineFireRedAsrModelConfig

SherpaOnnxOfflineFunASRNanoModelConfig

typedef struct SherpaOnnxOfflineFunASRNanoModelConfig SherpaOnnxOfflineFunASRNanoModelConfig

SherpaOnnxOfflineLMConfig

typedef struct SherpaOnnxOfflineLMConfig SherpaOnnxOfflineLMConfig

SherpaOnnxOfflineMedAsrCtcModelConfig

typedef struct SherpaOnnxOfflineMedAsrCtcModelConfig SherpaOnnxOfflineMedAsrCtcModelConfig

SherpaOnnxOfflineModelConfig

typedef struct SherpaOnnxOfflineModelConfig SherpaOnnxOfflineModelConfig

Zero-initialize this struct before use, then fill in exactly the sub-config needed by the model family you want to run.

Exactly one model family should be configured for each recognizer. For example, set only one of transducer, paraformer, nemo_ctc, whisper, tdnn, sense_voice, moonshine, fire_red_asr, dolphin, zipformer_ctc, canary, cohere_transcribe, wenet_ctc, omnilingual, medasr, funasr_nano, fire_red_asr_ctc, or qwen3_asr.

If multiple model families are configured at the same time, the implementation will choose one of them, and which one is used is implementation-defined. Do not rely on any precedence rule.

SherpaOnnxOfflineMoonshineModelConfig

typedef struct SherpaOnnxOfflineMoonshineModelConfig SherpaOnnxOfflineMoonshineModelConfig

SherpaOnnxOfflineNemoEncDecCtcModelConfig

typedef struct SherpaOnnxOfflineNemoEncDecCtcModelConfig SherpaOnnxOfflineNemoEncDecCtcModelConfig

SherpaOnnxOfflineOmnilingualAsrCtcModelConfig

typedef struct SherpaOnnxOfflineOmnilingualAsrCtcModelConfig SherpaOnnxOfflineOmnilingualAsrCtcModelConfig

SherpaOnnxOfflineParaformerModelConfig

typedef struct SherpaOnnxOfflineParaformerModelConfig SherpaOnnxOfflineParaformerModelConfig

SherpaOnnxOfflinePunctuation

typedef struct SherpaOnnxOfflinePunctuation SherpaOnnxOfflinePunctuation

Definition at line 3540 of file c-api.h.

SherpaOnnxOfflinePunctuationConfig

typedef struct SherpaOnnxOfflinePunctuationConfig SherpaOnnxOfflinePunctuationConfig

SherpaOnnxOfflinePunctuationModelConfig

typedef struct SherpaOnnxOfflinePunctuationModelConfig SherpaOnnxOfflinePunctuationModelConfig

Example:

SherpaOnnxOfflinePunctuationModelConfig model;
memset(&model, 0, sizeof(model));
model.ct_transformer =
    "./sherpa-onnx-punct-ct-transformer-zh-en-vocab272727-2024-04-12/model.onnx";
model.num_threads = 1;
model.provider = "cpu";

SherpaOnnxOfflineQwen3ASRModelConfig

typedef struct SherpaOnnxOfflineQwen3ASRModelConfig SherpaOnnxOfflineQwen3ASRModelConfig

SherpaOnnxOfflineRecognizer

typedef struct SherpaOnnxOfflineRecognizer SherpaOnnxOfflineRecognizer

Definition at line 1185 of file c-api.h.

SherpaOnnxOfflineRecognizerConfig

typedef struct SherpaOnnxOfflineRecognizerConfig SherpaOnnxOfflineRecognizerConfig

Zero-initialize this struct before use.

Example using Whisper:

SherpaOnnxOfflineRecognizerConfig config;
memset(&config, 0, sizeof(config));
 
config.feat_config.sample_rate = 16000;
config.feat_config.feature_dim = 80;
 
config.model_config.whisper.encoder =
    "./sherpa-onnx-whisper-tiny/tiny-encoder.onnx";
config.model_config.whisper.decoder =
    "./sherpa-onnx-whisper-tiny/tiny-decoder.onnx";
config.model_config.whisper.language = "en";
config.model_config.whisper.task = "transcribe";
config.model_config.tokens =
    "./sherpa-onnx-whisper-tiny/tiny-tokens.txt";
config.model_config.provider = "cpu";
config.model_config.num_threads = 1;
 
config.decoding_method = "greedy_search";

Example using SenseVoice:

config.model_config.sense_voice.model =
    "./sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17-int8/model.int8.onnx";
config.model_config.sense_voice.language = "auto";
config.model_config.sense_voice.use_itn = 1;
config.model_config.tokens =
    "./sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17-int8/tokens.txt";

Example using Parakeet TDT:

config.model_config.transducer.encoder =
    "./sherpa-onnx-nemo-parakeet-tdt-0.6b-v3-int8/encoder.int8.onnx";
config.model_config.transducer.decoder =
    "./sherpa-onnx-nemo-parakeet-tdt-0.6b-v3-int8/decoder.int8.onnx";
config.model_config.transducer.joiner =
    "./sherpa-onnx-nemo-parakeet-tdt-0.6b-v3-int8/joiner.int8.onnx";
config.model_config.tokens =
    "./sherpa-onnx-nemo-parakeet-tdt-0.6b-v3-int8/tokens.txt";
config.model_config.model_type = "nemo_transducer";

SherpaOnnxOfflineRecognizerResult

typedef struct SherpaOnnxOfflineRecognizerResult SherpaOnnxOfflineRecognizerResult

All pointers in this struct are owned by the result object returned from SherpaOnnxGetOfflineStreamResult() and become invalid after SherpaOnnxDestroyOfflineRecognizerResult() is called.

SherpaOnnxOfflineSenseVoiceModelConfig

typedef struct SherpaOnnxOfflineSenseVoiceModelConfig SherpaOnnxOfflineSenseVoiceModelConfig

SherpaOnnxOfflineSourceSeparation

typedef struct SherpaOnnxOfflineSourceSeparation SherpaOnnxOfflineSourceSeparation

Definition at line 4262 of file c-api.h.

SherpaOnnxOfflineSourceSeparationConfig

typedef struct SherpaOnnxOfflineSourceSeparationConfig SherpaOnnxOfflineSourceSeparationConfig

SherpaOnnxOfflineSourceSeparationModelConfig

typedef struct SherpaOnnxOfflineSourceSeparationModelConfig SherpaOnnxOfflineSourceSeparationModelConfig

SherpaOnnxOfflineSourceSeparationSpleeterModelConfig

typedef struct SherpaOnnxOfflineSourceSeparationSpleeterModelConfig SherpaOnnxOfflineSourceSeparationSpleeterModelConfig

SherpaOnnxOfflineSourceSeparationUvrModelConfig

typedef struct SherpaOnnxOfflineSourceSeparationUvrModelConfig SherpaOnnxOfflineSourceSeparationUvrModelConfig

SherpaOnnxOfflineSpeakerDiarization

typedef struct SherpaOnnxOfflineSpeakerDiarization SherpaOnnxOfflineSpeakerDiarization

Definition at line 3829 of file c-api.h.

SherpaOnnxOfflineSpeakerDiarizationConfig

typedef struct SherpaOnnxOfflineSpeakerDiarizationConfig SherpaOnnxOfflineSpeakerDiarizationConfig

Example based on offline-sepaker-diarization-c-api.c:

SherpaOnnxOfflineSpeakerDiarizationConfig config;
memset(&config, 0, sizeof(config));
config.segmentation.pyannote.model =
    "./sherpa-onnx-pyannote-segmentation-3-0/model.onnx";
config.embedding.model =
    "./3dspeaker_speech_eres2net_base_sv_zh-cn_3dspeaker_16k.onnx";
config.clustering.num_clusters = 4;

SherpaOnnxOfflineSpeakerDiarizationProgressCallback

typedef int32_t(* SherpaOnnxOfflineSpeakerDiarizationProgressCallback) (int32_t num_processed_chunks, int32_t num_total_chunks, void *arg)

The current implementation reports progress but ignores the callback's return value.

Definition at line 3938 of file c-api.h.

SherpaOnnxOfflineSpeakerDiarizationProgressCallbackNoArg

typedef int32_t(* SherpaOnnxOfflineSpeakerDiarizationProgressCallbackNoArg) (int32_t num_processed_chunks, int32_t num_total_chunks)

Definition at line 3945 of file c-api.h.

SherpaOnnxOfflineSpeakerDiarizationResult

typedef struct SherpaOnnxOfflineSpeakerDiarizationResult SherpaOnnxOfflineSpeakerDiarizationResult

Definition at line 3873 of file c-api.h.

SherpaOnnxOfflineSpeakerDiarizationSegment

typedef struct SherpaOnnxOfflineSpeakerDiarizationSegment SherpaOnnxOfflineSpeakerDiarizationSegment

SherpaOnnxOfflineSpeakerSegmentationModelConfig

typedef struct SherpaOnnxOfflineSpeakerSegmentationModelConfig SherpaOnnxOfflineSpeakerSegmentationModelConfig

Configure exactly one model family. If multiple model families are provided, one is chosen and the choice is implementation-defined.

SherpaOnnxOfflineSpeakerSegmentationPyannoteModelConfig

typedef struct SherpaOnnxOfflineSpeakerSegmentationPyannoteModelConfig SherpaOnnxOfflineSpeakerSegmentationPyannoteModelConfig

SherpaOnnxOfflineSpeechDenoiser

typedef struct SherpaOnnxOfflineSpeechDenoiser SherpaOnnxOfflineSpeechDenoiser

Definition at line 4053 of file c-api.h.

SherpaOnnxOfflineSpeechDenoiserConfig

typedef struct SherpaOnnxOfflineSpeechDenoiserConfig SherpaOnnxOfflineSpeechDenoiserConfig

SherpaOnnxOfflineSpeechDenoiserDpdfNetModelConfig

typedef struct SherpaOnnxOfflineSpeechDenoiserDpdfNetModelConfig SherpaOnnxOfflineSpeechDenoiserDpdfNetModelConfig

SherpaOnnxOfflineSpeechDenoiserGtcrnModelConfig

typedef struct SherpaOnnxOfflineSpeechDenoiserGtcrnModelConfig SherpaOnnxOfflineSpeechDenoiserGtcrnModelConfig

SherpaOnnxOfflineSpeechDenoiserModelConfig

typedef struct SherpaOnnxOfflineSpeechDenoiserModelConfig SherpaOnnxOfflineSpeechDenoiserModelConfig

Configure exactly one model family. If multiple model families are provided, one is chosen and the choice is implementation-defined.

SherpaOnnxOfflineStream

typedef struct SherpaOnnxOfflineStream SherpaOnnxOfflineStream

Definition at line 1188 of file c-api.h.

SherpaOnnxOfflineTdnnModelConfig

typedef struct SherpaOnnxOfflineTdnnModelConfig SherpaOnnxOfflineTdnnModelConfig

SherpaOnnxOfflineTransducerModelConfig

typedef struct SherpaOnnxOfflineTransducerModelConfig SherpaOnnxOfflineTransducerModelConfig

SherpaOnnxOfflineTts

typedef struct SherpaOnnxOfflineTts SherpaOnnxOfflineTts

Definition at line 2473 of file c-api.h.

SherpaOnnxOfflineTtsConfig

typedef struct SherpaOnnxOfflineTtsConfig SherpaOnnxOfflineTtsConfig

SherpaOnnxOfflineTtsConfig config;
memset(&config, 0, sizeof(config));
 
config.model.kokoro.model = "./kokoro-en-v0_19/model.onnx";
config.model.kokoro.voices = "./kokoro-en-v0_19/voices.bin";
config.model.kokoro.tokens = "./kokoro-en-v0_19/tokens.txt";
config.model.kokoro.data_dir = "./kokoro-en-v0_19/espeak-ng-data";
config.model.num_threads = 2;
config.model.provider = "cpu";
config.model.debug = 0;
config.max_num_sentences = 2;

SherpaOnnxOfflineTtsKittenModelConfig

typedef struct SherpaOnnxOfflineTtsKittenModelConfig SherpaOnnxOfflineTtsKittenModelConfig

SherpaOnnxOfflineTtsKokoroModelConfig

typedef struct SherpaOnnxOfflineTtsKokoroModelConfig SherpaOnnxOfflineTtsKokoroModelConfig

SherpaOnnxOfflineTtsMatchaModelConfig

typedef struct SherpaOnnxOfflineTtsMatchaModelConfig SherpaOnnxOfflineTtsMatchaModelConfig

SherpaOnnxOfflineTtsModelConfig

typedef struct SherpaOnnxOfflineTtsModelConfig SherpaOnnxOfflineTtsModelConfig

Exactly one TTS model family should be configured. For example, set only one of vits, matcha, kokoro, kitten, zipvoice, pocket, or supertonic.

If multiple model families are configured at the same time, the implementation will choose one of them, and which one is used is implementation-defined. Do not rely on any precedence rule.

Concrete example model packages in this repository include:

• kokoro-en-v0_19
• sherpa-onnx-pocket-tts-int8-2026-01-26
• matcha-icefall-en_US-ljspeech
• sherpa-onnx-zipvoice-distill-int8-zh-en-emilia

SherpaOnnxOfflineTtsPocketModelConfig

typedef struct SherpaOnnxOfflineTtsPocketModelConfig SherpaOnnxOfflineTtsPocketModelConfig

SherpaOnnxOfflineTtsSupertonicModelConfig

typedef struct SherpaOnnxOfflineTtsSupertonicModelConfig SherpaOnnxOfflineTtsSupertonicModelConfig

SherpaOnnxOfflineTtsVitsModelConfig

typedef struct SherpaOnnxOfflineTtsVitsModelConfig SherpaOnnxOfflineTtsVitsModelConfig

SherpaOnnxOfflineTtsZipvoiceModelConfig

typedef struct SherpaOnnxOfflineTtsZipvoiceModelConfig SherpaOnnxOfflineTtsZipvoiceModelConfig

SherpaOnnxOfflineWenetCtcModelConfig

typedef struct SherpaOnnxOfflineWenetCtcModelConfig SherpaOnnxOfflineWenetCtcModelConfig

SherpaOnnxOfflineWhisperModelConfig

typedef struct SherpaOnnxOfflineWhisperModelConfig SherpaOnnxOfflineWhisperModelConfig

SherpaOnnxOfflineZipformerAudioTaggingModelConfig

typedef struct SherpaOnnxOfflineZipformerAudioTaggingModelConfig SherpaOnnxOfflineZipformerAudioTaggingModelConfig

SherpaOnnxOfflineZipformerCtcModelConfig

typedef struct SherpaOnnxOfflineZipformerCtcModelConfig SherpaOnnxOfflineZipformerCtcModelConfig

SherpaOnnxOnlineCtcFstDecoderConfig

typedef struct SherpaOnnxOnlineCtcFstDecoderConfig SherpaOnnxOnlineCtcFstDecoderConfig

SherpaOnnxOnlineModelConfig

typedef struct SherpaOnnxOnlineModelConfig SherpaOnnxOnlineModelConfig

Zero-initialize this struct before use, then fill in the sub-config for the model family you want to use together with the shared fields such as tokens, provider, and num_threads.

Exactly one model family should be configured for each recognizer. For example, set only one of transducer, paraformer, zipformer2_ctc, nemo_ctc, or t_one_ctc.

If multiple model families are configured at the same time, the implementation will choose one of them, and which one is used is implementation-defined. Do not rely on any precedence rule.

SherpaOnnxOnlineNemoCtcModelConfig

typedef struct SherpaOnnxOnlineNemoCtcModelConfig SherpaOnnxOnlineNemoCtcModelConfig

SherpaOnnxOnlineParaformerModelConfig

typedef struct SherpaOnnxOnlineParaformerModelConfig SherpaOnnxOnlineParaformerModelConfig

Please visit https://k2-fsa.github.io/sherpa/onnx/pretrained_models/online-paraformer/index.html to download compatible models.

SherpaOnnxOnlinePunctuation

typedef struct SherpaOnnxOnlinePunctuation SherpaOnnxOnlinePunctuation

Definition at line 3613 of file c-api.h.

SherpaOnnxOnlinePunctuationConfig

typedef struct SherpaOnnxOnlinePunctuationConfig SherpaOnnxOnlinePunctuationConfig

SherpaOnnxOnlinePunctuationModelConfig

typedef struct SherpaOnnxOnlinePunctuationModelConfig SherpaOnnxOnlinePunctuationModelConfig

Example using sherpa-onnx-online-punct-en-2024-08-06:

SherpaOnnxOnlinePunctuationModelConfig model;
memset(&model, 0, sizeof(model));
model.cnn_bilstm =
"./sherpa-onnx-online-punct-en-2024-08-06/model.int8.onnx"; model.bpe_vocab =
"./sherpa-onnx-online-punct-en-2024-08-06/bpe.vocab"; model.num_threads = 1;
model.provider = "cpu";

SherpaOnnxOnlineRecognizer

typedef struct SherpaOnnxOnlineRecognizer SherpaOnnxOnlineRecognizer

Definition at line 422 of file c-api.h.

SherpaOnnxOnlineRecognizerConfig

typedef struct SherpaOnnxOnlineRecognizerConfig SherpaOnnxOnlineRecognizerConfig

Zero-initialize this struct before use. Then fill in feat_config, model_config, and any optional decoding, endpoint, or hotword settings.

Example model package: sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20

SherpaOnnxOnlineRecognizerConfig config;
memset(&config, 0, sizeof(config));
 
config.feat_config.sample_rate = 16000;
config.feat_config.feature_dim = 80;
 
config.model_config.transducer.encoder =
    "./sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/"
    "encoder-epoch-99-avg-1.int8.onnx";
config.model_config.transducer.decoder =
    "./sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/"
    "decoder-epoch-99-avg-1.onnx";
config.model_config.transducer.joiner =
    "./sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/"
    "joiner-epoch-99-avg-1.int8.onnx";
config.model_config.tokens =
    "./sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/"
    "tokens.txt";
config.model_config.provider = "cpu";
config.model_config.num_threads = 1;
 
config.decoding_method = "greedy_search";

SherpaOnnxOnlineRecognizerResult

typedef struct SherpaOnnxOnlineRecognizerResult SherpaOnnxOnlineRecognizerResult

All pointers in this struct are owned by the result object returned from SherpaOnnxGetOnlineStreamResult() and become invalid after SherpaOnnxDestroyOnlineRecognizerResult() is called.

SherpaOnnxOnlineSpeechDenoiser

typedef struct SherpaOnnxOnlineSpeechDenoiser SherpaOnnxOnlineSpeechDenoiser

Definition at line 4149 of file c-api.h.

SherpaOnnxOnlineSpeechDenoiserConfig

typedef struct SherpaOnnxOnlineSpeechDenoiserConfig SherpaOnnxOnlineSpeechDenoiserConfig

SherpaOnnxOnlineStream

typedef struct SherpaOnnxOnlineStream SherpaOnnxOnlineStream

Definition at line 424 of file c-api.h.

SherpaOnnxOnlineToneCtcModelConfig

typedef struct SherpaOnnxOnlineToneCtcModelConfig SherpaOnnxOnlineToneCtcModelConfig

SherpaOnnxOnlineTransducerModelConfig

typedef struct SherpaOnnxOnlineTransducerModelConfig SherpaOnnxOnlineTransducerModelConfig

Please refer to https://k2-fsa.github.io/sherpa/onnx/pretrained_models/index.html to download compatible pre-trained models.

SherpaOnnxOnlineZipformer2CtcModelConfig

typedef struct SherpaOnnxOnlineZipformer2CtcModelConfig SherpaOnnxOnlineZipformer2CtcModelConfig

SherpaOnnxResampleOut

typedef struct SherpaOnnxResampleOut SherpaOnnxResampleOut

Free this object with SherpaOnnxLinearResamplerResampleFree().

SherpaOnnxSileroVadModelConfig

typedef struct SherpaOnnxSileroVadModelConfig SherpaOnnxSileroVadModelConfig

SherpaOnnxSourceSeparationOutput

typedef struct SherpaOnnxSourceSeparationOutput SherpaOnnxSourceSeparationOutput

SherpaOnnxSourceSeparationStem

typedef struct SherpaOnnxSourceSeparationStem SherpaOnnxSourceSeparationStem

SherpaOnnxSpeakerEmbeddingExtractor

typedef struct SherpaOnnxSpeakerEmbeddingExtractor SherpaOnnxSpeakerEmbeddingExtractor

Definition at line 3054 of file c-api.h.

SherpaOnnxSpeakerEmbeddingExtractorConfig

typedef struct SherpaOnnxSpeakerEmbeddingExtractorConfig SherpaOnnxSpeakerEmbeddingExtractorConfig

Example using 3dspeaker_speech_campplus_sv_zh-cn_16k-common.onnx:

SherpaOnnxSpeakerEmbeddingExtractorConfig config;
memset(&config, 0, sizeof(config));
config.model = "./3dspeaker_speech_campplus_sv_zh-cn_16k-common.onnx";
config.num_threads = 1;
config.provider = "cpu";

SherpaOnnxSpeakerEmbeddingManager

typedef struct SherpaOnnxSpeakerEmbeddingManager SherpaOnnxSpeakerEmbeddingManager

Definition at line 3153 of file c-api.h.

SherpaOnnxSpeakerEmbeddingManagerBestMatchesResult

typedef struct SherpaOnnxSpeakerEmbeddingManagerBestMatchesResult SherpaOnnxSpeakerEmbeddingManagerBestMatchesResult

Free this object with SherpaOnnxSpeakerEmbeddingManagerFreeBestMatches().

SherpaOnnxSpeakerEmbeddingManagerSpeakerMatch

typedef struct SherpaOnnxSpeakerEmbeddingManagerSpeakerMatch SherpaOnnxSpeakerEmbeddingManagerSpeakerMatch

SherpaOnnxSpeechSegment

typedef struct SherpaOnnxSpeechSegment SherpaOnnxSpeechSegment

The segment owns samples. Free the whole object with SherpaOnnxDestroySpeechSegment().

SherpaOnnxSpokenLanguageIdentification

typedef struct SherpaOnnxSpokenLanguageIdentification SherpaOnnxSpokenLanguageIdentification

Definition at line 2937 of file c-api.h.

SherpaOnnxSpokenLanguageIdentificationConfig

typedef struct SherpaOnnxSpokenLanguageIdentificationConfig SherpaOnnxSpokenLanguageIdentificationConfig

The current implementation uses Whisper-based models.

Example using sherpa-onnx-whisper-tiny:

SherpaOnnxSpokenLanguageIdentificationConfig config;
memset(&config, 0, sizeof(config));
config.whisper.encoder = "./sherpa-onnx-whisper-tiny/tiny-encoder.int8.onnx";
config.whisper.decoder = "./sherpa-onnx-whisper-tiny/tiny-decoder.int8.onnx";
config.num_threads = 1;
config.provider = "cpu";

SherpaOnnxSpokenLanguageIdentificationResult

typedef struct SherpaOnnxSpokenLanguageIdentificationResult SherpaOnnxSpokenLanguageIdentificationResult

Free this object with SherpaOnnxDestroySpokenLanguageIdentificationResult().

SherpaOnnxSpokenLanguageIdentificationWhisperConfig

typedef struct SherpaOnnxSpokenLanguageIdentificationWhisperConfig SherpaOnnxSpokenLanguageIdentificationWhisperConfig

Example:

SherpaOnnxSpokenLanguageIdentificationWhisperConfig whisper;
memset(&whisper, 0, sizeof(whisper));
whisper.encoder = "./sherpa-onnx-whisper-tiny/tiny-encoder.int8.onnx";
whisper.decoder = "./sherpa-onnx-whisper-tiny/tiny-decoder.int8.onnx";

SherpaOnnxTenVadModelConfig

typedef struct SherpaOnnxTenVadModelConfig SherpaOnnxTenVadModelConfig

SherpaOnnxVadModelConfig

typedef struct SherpaOnnxVadModelConfig SherpaOnnxVadModelConfig

Exactly one VAD model family should be configured. Set either silero_vad.model or ten_vad.model.

If both are configured, the implementation will choose one of them, and which one is used is implementation-defined. Do not rely on any precedence rule.

Example model files:

• ./silero_vad.onnx
• ./ten-vad.onnx

SherpaOnnxVadModelConfig config;
memset(&config, 0, sizeof(config));
 
config.silero_vad.model = "./silero_vad.onnx";
config.silero_vad.threshold = 0.25f;
config.silero_vad.min_silence_duration = 0.5f;
config.silero_vad.min_speech_duration = 0.5f;
config.silero_vad.max_speech_duration = 10.0f;
config.silero_vad.window_size = 512;
 
config.sample_rate = 16000;
config.num_threads = 1;
config.provider = "cpu";
config.debug = 0;

SherpaOnnxVoiceActivityDetector

typedef struct SherpaOnnxVoiceActivityDetector SherpaOnnxVoiceActivityDetector

Definition at line 2054 of file c-api.h.

SherpaOnnxWave

typedef struct SherpaOnnxWave SherpaOnnxWave

Free this object with SherpaOnnxFreeWave().

Function Documentation

SherpaOfflinePunctuationAddPunct()

const char * SherpaOfflinePunctuationAddPunct	(	const SherpaOnnxOfflinePunctuation *	punct,
		const char *	text
	)

Parameters

punct	A pointer returned by SherpaOnnxCreateOfflinePunctuation().
text	Input text without punctuation.

Returns

A newly allocated punctuated string. Free it with SherpaOfflinePunctuationFreeText().

SherpaOfflinePunctuationFreeText()

void SherpaOfflinePunctuationFreeText

(

const char *

text

)

Parameters

text	A pointer returned by SherpaOfflinePunctuationAddPunct().

SherpaOnnxAcceptWaveformOffline()

void SherpaOnnxAcceptWaveformOffline	(	const SherpaOnnxOfflineStream *	stream,
		int32_t	sample_rate,
		const float *	samples,
		int32_t	n
	)

The input is mono floating-point PCM normalized to the range [-1, 1]. If sample_rate differs from the recognizer feature sample rate, sherpa-onnx resamples internally.

Warning

Call this function at most once for each offline stream. Offline recognition expects the entire utterance in a single call.

Parameters

stream	A pointer returned by SherpaOnnxCreateOfflineStream().
sample_rate	Sample rate of samples.
samples	Pointer to n samples in the range [-1, 1].
n	Number of samples.

const SherpaOnnxWave *wave =
    SherpaOnnxReadWave("./sherpa-onnx-whisper-tiny.en/test_wavs/0.wav");
const SherpaOnnxOfflineStream *stream =
    SherpaOnnxCreateOfflineStream(recognizer);
SherpaOnnxAcceptWaveformOffline(stream, wave->sample_rate,
                                wave->samples, wave->num_samples);
SherpaOnnxDecodeOfflineStream(recognizer, stream);

SherpaOnnxAudioTaggingCompute()

const SherpaOnnxAudioEvent *const * SherpaOnnxAudioTaggingCompute	(	const SherpaOnnxAudioTagging *	tagger,
		const SherpaOnnxOfflineStream *	s,
		int32_t	top_k
	)

The returned array is NULL-terminated. If top_k is -1, the value stored in config.top_k is used instead.

Parameters

tagger	A pointer returned by SherpaOnnxCreateAudioTagging().
s	A pointer returned by SherpaOnnxAudioTaggingCreateOfflineStream().
top_k	Number of top results to return, or -1 to use the configured default.

Returns

A newly allocated NULL-terminated array of result pointers ordered by descending probability. Free it with SherpaOnnxAudioTaggingFreeResults().

const SherpaOnnxAudioEvent *const *results =
    SherpaOnnxAudioTaggingCompute(tagger, stream, 5);
for (int32_t i = 0; results[i] != NULL; ++i) {
  printf("%d %.3f %s\n", results[i]->index, results[i]->prob,
         results[i]->name);
}
SherpaOnnxAudioTaggingFreeResults(results);

SherpaOnnxAudioTaggingCreateOfflineStream()

const SherpaOnnxOfflineStream * SherpaOnnxAudioTaggingCreateOfflineStream

(

const SherpaOnnxAudioTagging *

tagger

)

Parameters

tagger

A pointer returned by SherpaOnnxCreateAudioTagging().

Returns

A newly allocated offline stream. Free it with SherpaOnnxDestroyOfflineStream().

SherpaOnnxAudioTaggingFreeResults()

void SherpaOnnxAudioTaggingFreeResults

(

const SherpaOnnxAudioEvent *const *

p

)

Parameters

p	A pointer returned by SherpaOnnxAudioTaggingCompute().

SherpaOnnxCircularBufferFree()

void SherpaOnnxCircularBufferFree

(

const float *

p

)

SherpaOnnxCircularBufferGet()

const float * SherpaOnnxCircularBufferGet	(	const SherpaOnnxCircularBuffer *	buffer,
		int32_t	start_index,
		int32_t	n
	)

Parameters

buffer	A pointer returned by SherpaOnnxCreateCircularBuffer().
start_index	Absolute start index in the buffer timeline.
n	Number of samples to copy.

Returns

A newly allocated array containing n samples. Free it with SherpaOnnxCircularBufferFree().

const float *samples = SherpaOnnxCircularBufferGet(buffer, start, 3200);
SherpaOnnxCircularBufferFree(samples);

SherpaOnnxCircularBufferHead()

int32_t SherpaOnnxCircularBufferHead

(

const SherpaOnnxCircularBuffer *

buffer

)

The value is monotonically non-decreasing until SherpaOnnxCircularBufferReset() is called.

Parameters

buffer

A pointer returned by SherpaOnnxCreateCircularBuffer().

Returns

The current head index.

SherpaOnnxCircularBufferPop()

void SherpaOnnxCircularBufferPop	(	const SherpaOnnxCircularBuffer *	buffer,
		int32_t	n
	)

Parameters

buffer	A pointer returned by SherpaOnnxCreateCircularBuffer().
n	Number of samples to remove.

SherpaOnnxCircularBufferPush()

void SherpaOnnxCircularBufferPush	(	const SherpaOnnxCircularBuffer *	buffer,
		const float *	p,
		int32_t	n
	)

Parameters

buffer	A pointer returned by SherpaOnnxCreateCircularBuffer().
p	Pointer to n samples.
n	Number of samples.

SherpaOnnxCircularBufferPush(buffer, wave->samples, wave->num_samples);

SherpaOnnxCircularBufferReset()

void SherpaOnnxCircularBufferReset

(

const SherpaOnnxCircularBuffer *

buffer

)

Parameters

buffer

A pointer returned by SherpaOnnxCreateCircularBuffer().

SherpaOnnxCircularBufferSize()

int32_t SherpaOnnxCircularBufferSize

(

const SherpaOnnxCircularBuffer *

buffer

)

Parameters

buffer

A pointer returned by SherpaOnnxCreateCircularBuffer().

Returns

Number of samples currently in the buffer.

SherpaOnnxCreateAudioTagging()

const SherpaOnnxAudioTagging * SherpaOnnxCreateAudioTagging

(

const SherpaOnnxAudioTaggingConfig *

config

)

Parameters

config

Audio-tagging configuration.

Returns

A newly allocated audio tagger on success, or NULL on error. Free it with SherpaOnnxDestroyAudioTagging().

SherpaOnnxCreateCircularBuffer()

const SherpaOnnxCircularBuffer * SherpaOnnxCreateCircularBuffer

(

int32_t

capacity

)

Parameters

capacity

Maximum number of samples the buffer can keep.

Returns

A newly allocated buffer. Free it with SherpaOnnxDestroyCircularBuffer().

const SherpaOnnxCircularBuffer *buffer =
    SherpaOnnxCreateCircularBuffer(16000 * 30);

SherpaOnnxCreateDisplay()

const SherpaOnnxDisplay * SherpaOnnxCreateDisplay

(

int32_t

max_word_per_line

)

Parameters

max_word_per_line

Maximum number of words to show per line.

Returns

A newly allocated display helper. Free it with SherpaOnnxDestroyDisplay().

const SherpaOnnxDisplay *display = SherpaOnnxCreateDisplay(50);

SherpaOnnxCreateKeywordSpotter()

const SherpaOnnxKeywordSpotter * SherpaOnnxCreateKeywordSpotter

(

const SherpaOnnxKeywordSpotterConfig *

config

)

Parameters

config

Keyword spotter configuration.

Returns

A newly allocated keyword spotter on success, or NULL on error. Free it with SherpaOnnxDestroyKeywordSpotter().

SherpaOnnxCreateKeywordStream()

const SherpaOnnxOnlineStream * SherpaOnnxCreateKeywordStream

(

const SherpaOnnxKeywordSpotter *

spotter

)

Parameters

spotter

A pointer returned by SherpaOnnxCreateKeywordSpotter().

Returns

A newly allocated stream. Free it with SherpaOnnxDestroyOnlineStream().

SherpaOnnxCreateKeywordStreamWithKeywords()

const SherpaOnnxOnlineStream * SherpaOnnxCreateKeywordStreamWithKeywords	(	const SherpaOnnxKeywordSpotter *	spotter,
		const char *	keywords
	)

The keywords string uses the same textual format as the keyword files used by the examples. For instance:

const SherpaOnnxOnlineStream *stream =
    SherpaOnnxCreateKeywordStreamWithKeywords(
        kws, "y ǎn y uán @演员/zh ī m íng @知名");

Parameters

spotter	A pointer returned by SherpaOnnxCreateKeywordSpotter().
keywords	Inline keyword definition string.

Returns

A newly allocated stream. Free it with SherpaOnnxDestroyOnlineStream().

SherpaOnnxCreateLinearResampler()

const SherpaOnnxLinearResampler * SherpaOnnxCreateLinearResampler	(	int32_t	samp_rate_in_hz,
		int32_t	samp_rate_out_hz,
		float	filter_cutoff_hz,
		int32_t	num_zeros
	)

A common choice is:

float min_freq = samp_rate_in_hz < samp_rate_out_hz ? samp_rate_in_hz
                                                : samp_rate_out_hz;
float filter_cutoff_hz = 0.99f * 0.5f * min_freq;
int32_t num_zeros = 6;

Parameters

samp_rate_in_hz	Input sample rate in Hz.
samp_rate_out_hz	Output sample rate in Hz.
filter_cutoff_hz	Low-pass cutoff frequency in Hz.
num_zeros	Low-pass filter width control parameter.

Returns

A newly allocated resampler. Free it with SherpaOnnxDestroyLinearResampler().

SherpaOnnxCreateOfflinePunctuation()

const SherpaOnnxOfflinePunctuation * SherpaOnnxCreateOfflinePunctuation

(

const SherpaOnnxOfflinePunctuationConfig *

config

)

Parameters

config

Offline punctuation configuration.

Returns

A newly allocated punctuation processor on success, or NULL on error. Free it with SherpaOnnxDestroyOfflinePunctuation().

SherpaOnnxCreateOfflineRecognizer()

const SherpaOnnxOfflineRecognizer * SherpaOnnxCreateOfflineRecognizer

(

const SherpaOnnxOfflineRecognizerConfig *

config

)

Parameters

config

Recognizer configuration.

Returns

A recognizer handle on success, or NULL if the configuration is invalid. The caller owns the returned object and must free it with SherpaOnnxDestroyOfflineRecognizer().

Whisper example:

SherpaOnnxOfflineRecognizerConfig config;
memset(&config, 0, sizeof(config));
config.feat_config.sample_rate = 16000;
config.feat_config.feature_dim = 80;
config.model_config.whisper.encoder =
    "./sherpa-onnx-whisper-tiny/tiny-encoder.onnx";
config.model_config.whisper.decoder =
    "./sherpa-onnx-whisper-tiny/tiny-decoder.onnx";
config.model_config.whisper.language = "en";
config.model_config.whisper.task = "transcribe";
config.model_config.tokens =
    "./sherpa-onnx-whisper-tiny/tiny-tokens.txt";
config.model_config.provider = "cpu";
config.model_config.num_threads = 1;
config.decoding_method = "greedy_search";
 
const SherpaOnnxOfflineRecognizer *recognizer =
    SherpaOnnxCreateOfflineRecognizer(&config);

SenseVoice example:

config.model_config.sense_voice.model =
    "./sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17-int8/model.int8.onnx";
config.model_config.sense_voice.language = "auto";
config.model_config.sense_voice.use_itn = 1;
config.model_config.tokens =
    "./sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17-int8/tokens.txt";

Parakeet TDT example:

config.model_config.transducer.encoder =
    "./sherpa-onnx-nemo-parakeet-tdt-0.6b-v3-int8/encoder.int8.onnx";
config.model_config.transducer.decoder =
    "./sherpa-onnx-nemo-parakeet-tdt-0.6b-v3-int8/decoder.int8.onnx";
config.model_config.transducer.joiner =
    "./sherpa-onnx-nemo-parakeet-tdt-0.6b-v3-int8/joiner.int8.onnx";
config.model_config.tokens =
    "./sherpa-onnx-nemo-parakeet-tdt-0.6b-v3-int8/tokens.txt";
config.model_config.model_type = "nemo_transducer";

SherpaOnnxCreateOfflineSourceSeparation()

const SherpaOnnxOfflineSourceSeparation * SherpaOnnxCreateOfflineSourceSeparation

(

const SherpaOnnxOfflineSourceSeparationConfig *

config

)

Parameters

config

Source-separation configuration.

Returns

A newly allocated engine on success, or NULL on error. Free it with SherpaOnnxDestroyOfflineSourceSeparation().

SherpaOnnxCreateOfflineSpeakerDiarization()

const SherpaOnnxOfflineSpeakerDiarization * SherpaOnnxCreateOfflineSpeakerDiarization

(

const SherpaOnnxOfflineSpeakerDiarizationConfig *

config

)

Parameters

config

Offline speaker diarization configuration.

Returns

A newly allocated diarizer on success, or NULL on error. Free it with SherpaOnnxDestroyOfflineSpeakerDiarization().

SherpaOnnxCreateOfflineSpeechDenoiser()

const SherpaOnnxOfflineSpeechDenoiser * SherpaOnnxCreateOfflineSpeechDenoiser

(

const SherpaOnnxOfflineSpeechDenoiserConfig *

config

)

Example using gtcrn_simple.onnx:

SherpaOnnxOfflineSpeechDenoiserConfig config;
memset(&config, 0, sizeof(config));
config.model.gtcrn.model = "./gtcrn_simple.onnx";

Parameters

config

Offline denoiser configuration.

Returns

A newly allocated denoiser on success, or NULL on error. Free it with SherpaOnnxDestroyOfflineSpeechDenoiser().

SherpaOnnxCreateOfflineStream()

const SherpaOnnxOfflineStream * SherpaOnnxCreateOfflineStream

(

const SherpaOnnxOfflineRecognizer *

recognizer

)

Parameters

recognizer

A pointer returned by SherpaOnnxCreateOfflineRecognizer().

Returns

A newly created stream. The caller owns the returned object and must free it with SherpaOnnxDestroyOfflineStream().

const SherpaOnnxWave *wave =
    SherpaOnnxReadWave("./sherpa-onnx-whisper-tiny.en/test_wavs/0.wav");
const SherpaOnnxOfflineStream *stream =
    SherpaOnnxCreateOfflineStream(recognizer);

SherpaOnnxCreateOfflineStreamWithHotwords()

const SherpaOnnxOfflineStream * SherpaOnnxCreateOfflineStreamWithHotwords	(	const SherpaOnnxOfflineRecognizer *	recognizer,
		const char *	hotwords
	)

Parameters

recognizer	A pointer returned by SherpaOnnxCreateOfflineRecognizer().
hotwords	Hotwords text to associate with the stream.

Returns

A newly created stream. The caller owns the returned object and must free it with SherpaOnnxDestroyOfflineStream().

const SherpaOnnxOfflineStream *stream =
    SherpaOnnxCreateOfflineStreamWithHotwords(recognizer,
                                              "▁HELLO ▁WORLD");

SherpaOnnxCreateOfflineTts()

const SherpaOnnxOfflineTts * SherpaOnnxCreateOfflineTts

(

const SherpaOnnxOfflineTtsConfig *

config

)

Parameters

config

TTS configuration.

Returns

A newly allocated TTS engine on success, or NULL on configuration error. Free it with SherpaOnnxDestroyOfflineTts().

SherpaOnnxOfflineTtsConfig config;
memset(&config, 0, sizeof(config));
config.model.kokoro.model = "./kokoro-en-v0_19/model.onnx";
config.model.kokoro.voices = "./kokoro-en-v0_19/voices.bin";
config.model.kokoro.tokens = "./kokoro-en-v0_19/tokens.txt";
config.model.kokoro.data_dir = "./kokoro-en-v0_19/espeak-ng-data";
config.model.num_threads = 2;
 
const SherpaOnnxOfflineTts *tts = SherpaOnnxCreateOfflineTts(&config);

SherpaOnnxCreateOnlinePunctuation()

const SherpaOnnxOnlinePunctuation * SherpaOnnxCreateOnlinePunctuation

(

const SherpaOnnxOnlinePunctuationConfig *

config

)

Parameters

config

Online punctuation configuration.

Returns

A newly allocated punctuation processor on success, or NULL on error. Free it with SherpaOnnxDestroyOnlinePunctuation().

SherpaOnnxCreateOnlineRecognizer()

const SherpaOnnxOnlineRecognizer * SherpaOnnxCreateOnlineRecognizer

(

const SherpaOnnxOnlineRecognizerConfig *

config

)

The returned recognizer runs locally and does not require Internet access.

Parameters

config

Recognizer configuration.

Returns

A recognizer handle on success, or NULL if the configuration is invalid. The caller owns the returned object and must free it with SherpaOnnxDestroyOnlineRecognizer().

SherpaOnnxOnlineRecognizerConfig config;
memset(&config, 0, sizeof(config));
config.feat_config.sample_rate = 16000;
config.feat_config.feature_dim = 80;
config.model_config.transducer.encoder =
    "./sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/"
    "encoder-epoch-99-avg-1.int8.onnx";
config.model_config.transducer.decoder =
    "./sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/"
    "decoder-epoch-99-avg-1.onnx";
config.model_config.transducer.joiner =
    "./sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/"
    "joiner-epoch-99-avg-1.int8.onnx";
config.model_config.tokens =
    "./sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/"
    "tokens.txt";
config.model_config.provider = "cpu";
config.model_config.num_threads = 1;
config.decoding_method = "greedy_search";
 
const SherpaOnnxOnlineRecognizer *recognizer =
    SherpaOnnxCreateOnlineRecognizer(&config);

SherpaOnnxCreateOnlineSpeechDenoiser()

const SherpaOnnxOnlineSpeechDenoiser * SherpaOnnxCreateOnlineSpeechDenoiser

(

const SherpaOnnxOnlineSpeechDenoiserConfig *

config

)

Parameters

config

Online denoiser configuration.

Returns

A newly allocated denoiser on success, or NULL on error. Free it with SherpaOnnxDestroyOnlineSpeechDenoiser().

SherpaOnnxCreateOnlineStream()

const SherpaOnnxOnlineStream * SherpaOnnxCreateOnlineStream

(

const SherpaOnnxOnlineRecognizer *

recognizer

)

One stream corresponds to one decoding state. Reuse the same recognizer to create multiple streams.

Parameters

recognizer

A pointer returned by SherpaOnnxCreateOnlineRecognizer().

Returns

A newly created stream. The caller owns the returned object and must free it with SherpaOnnxDestroyOnlineStream().

const SherpaOnnxWave *wave = SherpaOnnxReadWave(
    "./sherpa-onnx-streaming-paraformer-bilingual-zh-en/test_wavs/0.wav");
const SherpaOnnxOnlineStream *stream =
    SherpaOnnxCreateOnlineStream(recognizer);

SherpaOnnxCreateOnlineStreamWithHotwords()

const SherpaOnnxOnlineStream * SherpaOnnxCreateOnlineStreamWithHotwords	(	const SherpaOnnxOnlineRecognizer *	recognizer,
		const char *	hotwords
	)

Parameters

recognizer	A pointer returned by SherpaOnnxCreateOnlineRecognizer().
hotwords	Hotwords text to associate with the stream.

Returns

A newly created stream. The caller owns the returned object and must free it with SherpaOnnxDestroyOnlineStream().

const SherpaOnnxOnlineStream *stream =
    SherpaOnnxCreateOnlineStreamWithHotwords(recognizer,
                                             "▁HELLO ▁WORLD");

SherpaOnnxCreateSpeakerEmbeddingExtractor()

const SherpaOnnxSpeakerEmbeddingExtractor * SherpaOnnxCreateSpeakerEmbeddingExtractor

(

const SherpaOnnxSpeakerEmbeddingExtractorConfig *

config

)

Parameters

config

Speaker embedding extractor configuration.

Returns

A newly allocated extractor on success, or NULL on error. Free it with SherpaOnnxDestroySpeakerEmbeddingExtractor().

SherpaOnnxCreateSpeakerEmbeddingManager()

const SherpaOnnxSpeakerEmbeddingManager * SherpaOnnxCreateSpeakerEmbeddingManager

(

int32_t

dim

)

The manager stores enrolled speaker embeddings and supports speaker search and verification.

Parameters

dim	Embedding dimension. This should match SherpaOnnxSpeakerEmbeddingExtractorDim().

Returns

A newly allocated manager. Free it with SherpaOnnxDestroySpeakerEmbeddingManager().

SherpaOnnxCreateSpokenLanguageIdentification()

const SherpaOnnxSpokenLanguageIdentification * SherpaOnnxCreateSpokenLanguageIdentification

(

const SherpaOnnxSpokenLanguageIdentificationConfig *

config

)

Parameters

config

Spoken-language identification configuration.

Returns

A newly allocated identifier on success, or NULL on error. Free it with SherpaOnnxDestroySpokenLanguageIdentification().

SherpaOnnxCreateVoiceActivityDetector()

const SherpaOnnxVoiceActivityDetector * SherpaOnnxCreateVoiceActivityDetector	(	const SherpaOnnxVadModelConfig *	config,
		float	buffer_size_in_seconds
	)

Example model files are shown in c-api-examples/vad-whisper-c-api.c.

Parameters

config	VAD configuration.
buffer_size_in_seconds	Internal buffering capacity in seconds.

Returns

A newly allocated detector on success, or NULL on configuration error. Free it with SherpaOnnxDestroyVoiceActivityDetector().

SherpaOnnxVadModelConfig config;
memset(&config, 0, sizeof(config));
config.silero_vad.model = "./silero_vad.onnx";
config.silero_vad.threshold = 0.25f;
config.silero_vad.min_silence_duration = 0.5f;
config.silero_vad.min_speech_duration = 0.5f;
config.silero_vad.max_speech_duration = 10.0f;
config.silero_vad.window_size = 512;
config.sample_rate = 16000;
config.num_threads = 1;
 
const SherpaOnnxVoiceActivityDetector *vad =
    SherpaOnnxCreateVoiceActivityDetector(&config, 30.0f);

SherpaOnnxDecodeKeywordStream()

void SherpaOnnxDecodeKeywordStream	(	const SherpaOnnxKeywordSpotter *	spotter,
		const SherpaOnnxOnlineStream *	stream
	)

Call this only when SherpaOnnxIsKeywordStreamReady() returns 1.

Parameters

spotter	A pointer returned by SherpaOnnxCreateKeywordSpotter().
stream	A pointer returned by SherpaOnnxCreateKeywordStream() or SherpaOnnxCreateKeywordStreamWithKeywords().

SherpaOnnxDecodeMultipleKeywordStreams()

void SherpaOnnxDecodeMultipleKeywordStreams	(	const SherpaOnnxKeywordSpotter *	spotter,
		const SherpaOnnxOnlineStream **	streams,
		int32_t	n
	)

The caller must ensure every stream in streams is ready before calling this function.

Parameters

spotter	A pointer returned by SherpaOnnxCreateKeywordSpotter().
streams	Array of ready streams.
n	Number of elements in streams.

SherpaOnnxDecodeMultipleOfflineStreams()

void SherpaOnnxDecodeMultipleOfflineStreams	(	const SherpaOnnxOfflineRecognizer *	recognizer,
		const SherpaOnnxOfflineStream **	streams,
		int32_t	n
	)

The caller must have already provided one utterance to each stream via SherpaOnnxAcceptWaveformOffline().

Parameters

recognizer	A pointer returned by SherpaOnnxCreateOfflineRecognizer().
streams	Array of n offline stream pointers.
n	Number of streams in streams.

const SherpaOnnxOfflineStream *streams[2] = {stream1, stream2};
SherpaOnnxDecodeMultipleOfflineStreams(recognizer, streams, 2);

SherpaOnnxDecodeMultipleOnlineStreams()

void SherpaOnnxDecodeMultipleOnlineStreams	(	const SherpaOnnxOnlineRecognizer *	recognizer,
		const SherpaOnnxOnlineStream **	streams,
		int32_t	n
	)

The caller must ensure every stream in streams is ready before calling this function.

Parameters

recognizer	A pointer returned by SherpaOnnxCreateOnlineRecognizer().
streams	Array of n stream pointers.
n	Number of streams in streams.

const SherpaOnnxOnlineStream *streams[2] = {stream1, stream2};
SherpaOnnxDecodeMultipleOnlineStreams(recognizer, streams, 2);

SherpaOnnxDecodeOfflineStream()

void SherpaOnnxDecodeOfflineStream	(	const SherpaOnnxOfflineRecognizer *	recognizer,
		const SherpaOnnxOfflineStream *	stream
	)

Call this after SherpaOnnxAcceptWaveformOffline().

Parameters

recognizer	A pointer returned by SherpaOnnxCreateOfflineRecognizer().
stream	A pointer returned by SherpaOnnxCreateOfflineStream().

SherpaOnnxDecodeOfflineStream(recognizer, stream);

SherpaOnnxDecodeOnlineStream()

void SherpaOnnxDecodeOnlineStream	(	const SherpaOnnxOnlineRecognizer *	recognizer,
		const SherpaOnnxOnlineStream *	stream
	)

Call this only when SherpaOnnxIsOnlineStreamReady() returns 1.

Parameters

recognizer	A pointer returned by SherpaOnnxCreateOnlineRecognizer().
stream	A pointer returned by SherpaOnnxCreateOnlineStream().

SherpaOnnxOnlineStreamAcceptWaveform(stream, sample_rate, samples, n);
while (SherpaOnnxIsOnlineStreamReady(recognizer, stream)) {
  SherpaOnnxDecodeOnlineStream(recognizer, stream);
}

SherpaOnnxDestroyAudioTagging()

void SherpaOnnxDestroyAudioTagging

(

const SherpaOnnxAudioTagging *

tagger

)

Parameters

tagger

A pointer returned by SherpaOnnxCreateAudioTagging().

SherpaOnnxDestroyCircularBuffer()

void SherpaOnnxDestroyCircularBuffer

(

const SherpaOnnxCircularBuffer *

buffer

)

Parameters

buffer

A pointer returned by SherpaOnnxCreateCircularBuffer().

SherpaOnnxDestroyCircularBuffer(buffer);
buffer = NULL;

SherpaOnnxDestroyDenoisedAudio()

void SherpaOnnxDestroyDenoisedAudio

(

const SherpaOnnxDenoisedAudio *

p

)

Parameters

p	A pointer returned by SherpaOnnxOfflineSpeechDenoiserRun(), SherpaOnnxOnlineSpeechDenoiserRun(), or SherpaOnnxOnlineSpeechDenoiserFlush().

SherpaOnnxDestroyDisplay()

void SherpaOnnxDestroyDisplay

(

const SherpaOnnxDisplay *

display

)

Parameters

display

A pointer returned by SherpaOnnxCreateDisplay().

SherpaOnnxDestroyKeywordResult()

void SherpaOnnxDestroyKeywordResult

(

const SherpaOnnxKeywordResult *

r

)

Parameters

r	A pointer returned by SherpaOnnxGetKeywordResult().

SherpaOnnxDestroyKeywordSpotter()

void SherpaOnnxDestroyKeywordSpotter

(

const SherpaOnnxKeywordSpotter *

spotter

)

Parameters

spotter

A pointer returned by SherpaOnnxCreateKeywordSpotter().

SherpaOnnxDestroyLinearResampler()

void SherpaOnnxDestroyLinearResampler

(

const SherpaOnnxLinearResampler *

p

)

Parameters

p	A pointer returned by SherpaOnnxCreateLinearResampler().

SherpaOnnxDestroyOfflinePunctuation()

void SherpaOnnxDestroyOfflinePunctuation

(

const SherpaOnnxOfflinePunctuation *

punct

)

Parameters

punct

A pointer returned by SherpaOnnxCreateOfflinePunctuation().

SherpaOnnxDestroyOfflineRecognizer()

void SherpaOnnxDestroyOfflineRecognizer

(

const SherpaOnnxOfflineRecognizer *

recognizer

)

Parameters

recognizer

A pointer returned by SherpaOnnxCreateOfflineRecognizer().

SherpaOnnxDestroyOfflineRecognizer(recognizer);
recognizer = NULL;

SherpaOnnxDestroyOfflineRecognizerResult()

void SherpaOnnxDestroyOfflineRecognizerResult

(

const SherpaOnnxOfflineRecognizerResult *

r

)

Parameters

r	A pointer returned by SherpaOnnxGetOfflineStreamResult().

SherpaOnnxDestroyOfflineRecognizerResult(r);
r = NULL;

SherpaOnnxDestroyOfflineSourceSeparation()

void SherpaOnnxDestroyOfflineSourceSeparation

(

const SherpaOnnxOfflineSourceSeparation *

ss

)

Parameters

ss	A pointer returned by SherpaOnnxCreateOfflineSourceSeparation().

SherpaOnnxDestroyOfflineSpeakerDiarization()

void SherpaOnnxDestroyOfflineSpeakerDiarization

(

const SherpaOnnxOfflineSpeakerDiarization *

sd

)

Parameters

sd	A pointer returned by SherpaOnnxCreateOfflineSpeakerDiarization().

SherpaOnnxDestroyOfflineSpeechDenoiser()

void SherpaOnnxDestroyOfflineSpeechDenoiser

(

const SherpaOnnxOfflineSpeechDenoiser *

sd

)

Parameters

sd	A pointer returned by SherpaOnnxCreateOfflineSpeechDenoiser().

SherpaOnnxDestroyOfflineStream()

void SherpaOnnxDestroyOfflineStream

(

const SherpaOnnxOfflineStream *

stream

)

Parameters

stream

A pointer returned by SherpaOnnxCreateOfflineStream() or SherpaOnnxCreateOfflineStreamWithHotwords().

SherpaOnnxDestroyOfflineStream(stream);
stream = NULL;

SherpaOnnxDestroyOfflineStreamResultJson()

void SherpaOnnxDestroyOfflineStreamResultJson

(

const char *

s

)

Parameters

s	A pointer returned by SherpaOnnxGetOfflineStreamResultAsJson().

SherpaOnnxDestroyOfflineStreamResultJson(json);
json = NULL;

SherpaOnnxDestroyOfflineTts()

void SherpaOnnxDestroyOfflineTts

(

const SherpaOnnxOfflineTts *

tts

)

Parameters

tts	A pointer returned by SherpaOnnxCreateOfflineTts().

SherpaOnnxDestroyOfflineTtsGeneratedAudio()

void SherpaOnnxDestroyOfflineTtsGeneratedAudio

(

const SherpaOnnxGeneratedAudio *

p

)

Parameters

p	A pointer returned by one of the SherpaOnnxOfflineTtsGenerate* functions.

SherpaOnnxDestroyOnlinePunctuation()

void SherpaOnnxDestroyOnlinePunctuation

(

const SherpaOnnxOnlinePunctuation *

punctuation

)

Parameters

punctuation

A pointer returned by SherpaOnnxCreateOnlinePunctuation().

SherpaOnnxDestroyOnlineRecognizer()

void SherpaOnnxDestroyOnlineRecognizer

(

const SherpaOnnxOnlineRecognizer *

recognizer

)

Parameters

recognizer

A pointer returned by SherpaOnnxCreateOnlineRecognizer().

SherpaOnnxDestroyOnlineRecognizer(recognizer);
recognizer = NULL;

SherpaOnnxDestroyOnlineRecognizerResult()

void SherpaOnnxDestroyOnlineRecognizerResult

(

const SherpaOnnxOnlineRecognizerResult *

r

)

Parameters

r	A pointer returned by SherpaOnnxGetOnlineStreamResult().

SherpaOnnxDestroyOnlineRecognizerResult(r);
r = NULL;

SherpaOnnxDestroyOnlineSpeechDenoiser()

void SherpaOnnxDestroyOnlineSpeechDenoiser

(

const SherpaOnnxOnlineSpeechDenoiser *

sd

)

Parameters

sd	A pointer returned by SherpaOnnxCreateOnlineSpeechDenoiser().

SherpaOnnxDestroyOnlineStream()

void SherpaOnnxDestroyOnlineStream

(

const SherpaOnnxOnlineStream *

stream

)

Parameters

stream

A pointer returned by SherpaOnnxCreateOnlineStream() or SherpaOnnxCreateOnlineStreamWithHotwords().

SherpaOnnxDestroyOnlineStream(stream);
stream = NULL;

SherpaOnnxDestroyOnlineStreamResultJson()

void SherpaOnnxDestroyOnlineStreamResultJson

(

const char *

s

)

Parameters

s	A pointer returned by SherpaOnnxGetOnlineStreamResultAsJson().

SherpaOnnxDestroyOnlineStreamResultJson(json);
json = NULL;

SherpaOnnxDestroySourceSeparationOutput()

void SherpaOnnxDestroySourceSeparationOutput

(

const SherpaOnnxSourceSeparationOutput *

p

)

Parameters

p	A pointer returned by SherpaOnnxOfflineSourceSeparationProcess().

SherpaOnnxDestroySpeakerEmbeddingExtractor()

void SherpaOnnxDestroySpeakerEmbeddingExtractor

(

const SherpaOnnxSpeakerEmbeddingExtractor *

p

)

Parameters

p	A pointer returned by SherpaOnnxCreateSpeakerEmbeddingExtractor().

SherpaOnnxDestroySpeakerEmbeddingManager()

void SherpaOnnxDestroySpeakerEmbeddingManager

(

const SherpaOnnxSpeakerEmbeddingManager *

p

)

Parameters

p	A pointer returned by SherpaOnnxCreateSpeakerEmbeddingManager().

SherpaOnnxDestroySpeechSegment()

void SherpaOnnxDestroySpeechSegment

(

const SherpaOnnxSpeechSegment *

p

)

Parameters

p	A pointer returned by SherpaOnnxVoiceActivityDetectorFront().

SherpaOnnxDestroySpokenLanguageIdentification()

void SherpaOnnxDestroySpokenLanguageIdentification

(

const SherpaOnnxSpokenLanguageIdentification *

slid

)

Parameters

slid	A pointer returned by SherpaOnnxCreateSpokenLanguageIdentification().

SherpaOnnxDestroySpokenLanguageIdentificationResult()

void SherpaOnnxDestroySpokenLanguageIdentificationResult

(

const SherpaOnnxSpokenLanguageIdentificationResult *

r

)

Parameters

r	A pointer returned by SherpaOnnxSpokenLanguageIdentificationCompute().

SherpaOnnxDestroyVoiceActivityDetector()

void SherpaOnnxDestroyVoiceActivityDetector

(

const SherpaOnnxVoiceActivityDetector *

p

)

Parameters

p	A pointer returned by SherpaOnnxCreateVoiceActivityDetector().

SherpaOnnxFileExists()

int32_t SherpaOnnxFileExists

(

const char *

filename

)

Parameters

filename

File path to test.

Returns

1 if the file exists; otherwise 0.

if (!SherpaOnnxFileExists("./Obama.wav")) {
  fprintf(stderr, "Please download Obama.wav\n");
}

SherpaOnnxFreeKeywordResultJson()

void SherpaOnnxFreeKeywordResultJson

(

const char *

s

)

Parameters

s	A pointer returned by SherpaOnnxGetKeywordResultAsJson().

SherpaOnnxFreeMultiChannelWave()

void SherpaOnnxFreeMultiChannelWave

(

const SherpaOnnxMultiChannelWave *

wave

)

Parameters

wave	A pointer returned by SherpaOnnxReadWaveMultiChannel().

SherpaOnnxFreeWave()

void SherpaOnnxFreeWave

(

const SherpaOnnxWave *

wave

)

SherpaOnnxGetGitDate()

const char * SherpaOnnxGetGitDate

(

)

The returned pointer refers to statically allocated memory owned by the library. Do not free it and do not modify it.

Returns

Build date string, for example "Fri Jun 20 11:22:52 2025".

SherpaOnnxGetGitSha1()

const char * SherpaOnnxGetGitSha1

(

)

The returned pointer refers to statically allocated memory owned by the library. Do not free it and do not modify it.

Returns

Short Git SHA1 string, for example "6982b86c".

SherpaOnnxGetKeywordResult()

const SherpaOnnxKeywordResult * SherpaOnnxGetKeywordResult	(	const SherpaOnnxKeywordSpotter *	spotter,
		const SherpaOnnxOnlineStream *	stream
	)

The returned snapshot may represent either "no trigger yet" or a detected keyword. A common pattern is to check whether strlen(r->keyword) != 0.

Parameters

spotter	A pointer returned by SherpaOnnxCreateKeywordSpotter().
stream	A pointer returned by SherpaOnnxCreateKeywordStream() or SherpaOnnxCreateKeywordStreamWithKeywords().

Returns

A newly allocated result snapshot. Free it with SherpaOnnxDestroyKeywordResult().

const SherpaOnnxKeywordResult *r = SherpaOnnxGetKeywordResult(kws, stream);
if (r && r->json && strlen(r->keyword)) {
  fprintf(stderr, "Detected keyword: %s\n", r->json);
  SherpaOnnxResetKeywordStream(kws, stream);
}
SherpaOnnxDestroyKeywordResult(r);

SherpaOnnxGetKeywordResultAsJson()

const char * SherpaOnnxGetKeywordResultAsJson	(	const SherpaOnnxKeywordSpotter *	spotter,
		const SherpaOnnxOnlineStream *	stream
	)

Parameters

spotter	A pointer returned by SherpaOnnxCreateKeywordSpotter().
stream	A pointer returned by SherpaOnnxCreateKeywordStream() or SherpaOnnxCreateKeywordStreamWithKeywords().

Returns

A newly allocated JSON string. Free it with SherpaOnnxFreeKeywordResultJson().

SherpaOnnxGetOfflineStreamResult()

const SherpaOnnxOfflineRecognizerResult * SherpaOnnxGetOfflineStreamResult

(

const SherpaOnnxOfflineStream *

stream

)

Call this after SherpaOnnxDecodeOfflineStream() or SherpaOnnxDecodeMultipleOfflineStreams().

Parameters

stream

A pointer returned by SherpaOnnxCreateOfflineStream().

Returns

A newly allocated result snapshot. Free it with SherpaOnnxDestroyOfflineRecognizerResult().

const SherpaOnnxOfflineRecognizerResult *r =
    SherpaOnnxGetOfflineStreamResult(stream);
printf("%s\n", r->text);
if (r->timestamps) {
  printf("First token starts at %.3f seconds\n", r->timestamps[0]);
}
SherpaOnnxDestroyOfflineRecognizerResult(r);

SherpaOnnxGetOfflineStreamResultAsJson()

const char * SherpaOnnxGetOfflineStreamResultAsJson

(

const SherpaOnnxOfflineStream *

stream

)

Parameters

stream

A pointer returned by SherpaOnnxCreateOfflineStream().

Returns

A newly allocated JSON string. Free it with SherpaOnnxDestroyOfflineStreamResultJson().

const char *json = SherpaOnnxGetOfflineStreamResultAsJson(stream);
puts(json);
SherpaOnnxDestroyOfflineStreamResultJson(json);

SherpaOnnxGetOnlineStreamResult()

const SherpaOnnxOnlineRecognizerResult * SherpaOnnxGetOnlineStreamResult	(	const SherpaOnnxOnlineRecognizer *	recognizer,
		const SherpaOnnxOnlineStream *	stream
	)

The returned snapshot is independent from the stream state. The caller owns it and must free it with SherpaOnnxDestroyOnlineRecognizerResult().

Parameters

recognizer	A pointer returned by SherpaOnnxCreateOnlineRecognizer().
stream	A pointer returned by SherpaOnnxCreateOnlineStream().

Returns

A newly allocated result snapshot.

const SherpaOnnxOnlineRecognizerResult *r =
    SherpaOnnxGetOnlineStreamResult(recognizer, stream);
printf("%s\n", r->text);
// r->tokens_arr[i] and r->timestamps[i] are parallel when timestamps
// are available.
SherpaOnnxDestroyOnlineRecognizerResult(r);

SherpaOnnxGetOnlineStreamResultAsJson()

const char * SherpaOnnxGetOnlineStreamResultAsJson	(	const SherpaOnnxOnlineRecognizer *	recognizer,
		const SherpaOnnxOnlineStream *	stream
	)

Parameters

recognizer	A pointer returned by SherpaOnnxCreateOnlineRecognizer().
stream	A pointer returned by SherpaOnnxCreateOnlineStream().

Returns

A newly allocated JSON string. Free it with SherpaOnnxDestroyOnlineStreamResultJson().

const char *json =
    SherpaOnnxGetOnlineStreamResultAsJson(recognizer, stream);
puts(json);
SherpaOnnxDestroyOnlineStreamResultJson(json);

SherpaOnnxGetVersionStr()

const char * SherpaOnnxGetVersionStr

(

)

The returned pointer refers to statically allocated memory owned by the library. Do not free it and do not modify it.

Returns

Version string, for example "1.12.1".

printf("sherpa-onnx version: %s\n", SherpaOnnxGetVersionStr());

SherpaOnnxIsKeywordStreamReady()

int32_t SherpaOnnxIsKeywordStreamReady	(	const SherpaOnnxKeywordSpotter *	spotter,
		const SherpaOnnxOnlineStream *	stream
	)

Parameters

spotter	A pointer returned by SherpaOnnxCreateKeywordSpotter().
stream	A pointer returned by SherpaOnnxCreateKeywordStream() or SherpaOnnxCreateKeywordStreamWithKeywords().

Returns

1 if the stream is ready to decode; otherwise 0.

SherpaOnnxIsOnlineStreamReady()

int32_t SherpaOnnxIsOnlineStreamReady	(	const SherpaOnnxOnlineRecognizer *	recognizer,
		const SherpaOnnxOnlineStream *	stream
	)

Parameters

recognizer	A pointer returned by SherpaOnnxCreateOnlineRecognizer().
stream	A pointer returned by SherpaOnnxCreateOnlineStream().

Returns

1 if enough frames are available for decoding; otherwise 0.

if (SherpaOnnxIsOnlineStreamReady(recognizer, stream)) {
  SherpaOnnxDecodeOnlineStream(recognizer, stream);
}

SherpaOnnxLinearResamplerResample()

const SherpaOnnxResampleOut * SherpaOnnxLinearResamplerResample	(	const SherpaOnnxLinearResampler *	p,
		const float *	input,
		int32_t	input_dim,
		int32_t	flush
	)

Set flush to 1 for the final chunk so buffered samples are emitted.

Parameters

p	A pointer returned by SherpaOnnxCreateLinearResampler().
input	Input sample array.
input_dim	Number of input samples.
flush	1 if this is the final chunk; otherwise 0.

Returns

A newly allocated output chunk. Free it with SherpaOnnxLinearResamplerResampleFree().

SherpaOnnxLinearResamplerResampleFree()

void SherpaOnnxLinearResamplerResampleFree

(

const SherpaOnnxResampleOut *

p

)

Parameters

p	A pointer returned by SherpaOnnxLinearResamplerResample().

SherpaOnnxLinearResamplerResampleGetInputSampleRate()

int32_t SherpaOnnxLinearResamplerResampleGetInputSampleRate

(

const SherpaOnnxLinearResampler *

p

)

Parameters

p	A pointer returned by SherpaOnnxCreateLinearResampler().

Returns

Input sample rate in Hz.

SherpaOnnxLinearResamplerResampleGetOutputSampleRate()

int32_t SherpaOnnxLinearResamplerResampleGetOutputSampleRate

(

const SherpaOnnxLinearResampler *

p

)

Parameters

p	A pointer returned by SherpaOnnxCreateLinearResampler().

Returns

Output sample rate in Hz.

SherpaOnnxLinearResamplerReset()

void SherpaOnnxLinearResamplerReset

(

const SherpaOnnxLinearResampler *

p

)

Parameters

p	A pointer returned by SherpaOnnxCreateLinearResampler().

SherpaOnnxOfflineRecognizerSetConfig()

void SherpaOnnxOfflineRecognizerSetConfig	(	const SherpaOnnxOfflineRecognizer *	recognizer,
		const SherpaOnnxOfflineRecognizerConfig *	config
	)

Parameters

recognizer	Recognizer handle.
config	New recognizer configuration.

SherpaOnnxOfflineRecognizerSetConfig(recognizer, &config);

SherpaOnnxOfflineSourceSeparationGetNumberOfStems()

int32_t SherpaOnnxOfflineSourceSeparationGetNumberOfStems

(

const SherpaOnnxOfflineSourceSeparation *

ss

)

For Spleeter 2-stems this returns 2 (vocals + accompaniment).

Parameters

ss	A pointer returned by SherpaOnnxCreateOfflineSourceSeparation().

Returns

Number of output stems.

SherpaOnnxOfflineSourceSeparationGetOutputSampleRate()

int32_t SherpaOnnxOfflineSourceSeparationGetOutputSampleRate

(

const SherpaOnnxOfflineSourceSeparation *

ss

)

Parameters

ss	A pointer returned by SherpaOnnxCreateOfflineSourceSeparation().

Returns

Output sample rate in Hz.

SherpaOnnxOfflineSourceSeparationProcess()

const SherpaOnnxSourceSeparationOutput * SherpaOnnxOfflineSourceSeparationProcess	(	const SherpaOnnxOfflineSourceSeparation *	ss,
		const float *const *	samples,
		int32_t	num_channels,
		int32_t	num_samples,
		int32_t	sample_rate
	)

All input channels must have the same number of samples.

Parameters

ss	A pointer returned by SherpaOnnxCreateOfflineSourceSeparation().
samples	samples[c] is a float array for channel c, values in [-1, 1].
num_channels	Number of input channels.
num_samples	Number of samples per channel (all channels must have the same length).
sample_rate	Input sample rate in Hz.

Returns

A newly allocated output on success, or NULL on error. Free it with SherpaOnnxDestroySourceSeparationOutput().

SherpaOnnxOfflineSpeakerDiarizationDestroyResult()

void SherpaOnnxOfflineSpeakerDiarizationDestroyResult

(

const SherpaOnnxOfflineSpeakerDiarizationResult *

r

)

Parameters

r	A pointer returned by one of the SherpaOnnxOfflineSpeakerDiarizationProcess*() functions.

SherpaOnnxOfflineSpeakerDiarizationDestroySegment()

void SherpaOnnxOfflineSpeakerDiarizationDestroySegment

(

const SherpaOnnxOfflineSpeakerDiarizationSegment *

s

)

Parameters

s	A pointer returned by SherpaOnnxOfflineSpeakerDiarizationResultSortByStartTime().

SherpaOnnxOfflineSpeakerDiarizationGetSampleRate()

int32_t SherpaOnnxOfflineSpeakerDiarizationGetSampleRate

(

const SherpaOnnxOfflineSpeakerDiarization *

sd

)

Parameters

sd	A pointer returned by SherpaOnnxCreateOfflineSpeakerDiarization().

Returns

Required input sample rate in Hz.

SherpaOnnxOfflineSpeakerDiarizationProcess()

const SherpaOnnxOfflineSpeakerDiarizationResult * SherpaOnnxOfflineSpeakerDiarizationProcess	(	const SherpaOnnxOfflineSpeakerDiarization *	sd,
		const float *	samples,
		int32_t	n
	)

Parameters

sd	A pointer returned by SherpaOnnxCreateOfflineSpeakerDiarization().
samples	Input mono PCM samples normalized to [-1, 1].
n	Number of input samples.

Returns

A newly allocated diarization result. Free it with SherpaOnnxOfflineSpeakerDiarizationDestroyResult().

SherpaOnnxOfflineSpeakerDiarizationProcessWithCallback()

const SherpaOnnxOfflineSpeakerDiarizationResult * SherpaOnnxOfflineSpeakerDiarizationProcessWithCallback	(	const SherpaOnnxOfflineSpeakerDiarization *	sd,
		const float *	samples,
		int32_t	n,
		SherpaOnnxOfflineSpeakerDiarizationProgressCallback	callback,
		void *	arg
	)

Parameters

sd	A pointer returned by SherpaOnnxCreateOfflineSpeakerDiarization().
samples	Input mono PCM samples normalized to [-1, 1].
n	Number of input samples.
callback	Progress callback.
arg	User pointer forwarded to callback.

Returns

A newly allocated diarization result. Free it with SherpaOnnxOfflineSpeakerDiarizationDestroyResult().

static int32_t ProgressCallback(int32_t done, int32_t total, void *arg) {
  fprintf(stderr, "progress %.2f%%\n", 100.0f * done / total);
  return 0;
}

SherpaOnnxOfflineSpeakerDiarizationProcessWithCallbackNoArg()

const SherpaOnnxOfflineSpeakerDiarizationResult * SherpaOnnxOfflineSpeakerDiarizationProcessWithCallbackNoArg	(	const SherpaOnnxOfflineSpeakerDiarization *	sd,
		const float *	samples,
		int32_t	n,
		SherpaOnnxOfflineSpeakerDiarizationProgressCallbackNoArg	callback
	)

Parameters

sd	A pointer returned by SherpaOnnxCreateOfflineSpeakerDiarization().
samples	Input mono PCM samples normalized to [-1, 1].
n	Number of input samples.
callback	Progress callback.

Returns

A newly allocated diarization result. Free it with SherpaOnnxOfflineSpeakerDiarizationDestroyResult().

SherpaOnnxOfflineSpeakerDiarizationResultGetNumSegments()

int32_t SherpaOnnxOfflineSpeakerDiarizationResultGetNumSegments

(

const SherpaOnnxOfflineSpeakerDiarizationResult *

r

)

Parameters

r	A pointer returned by one of the SherpaOnnxOfflineSpeakerDiarizationProcess*() functions.

Returns

Number of segments.

SherpaOnnxOfflineSpeakerDiarizationResultGetNumSpeakers()

int32_t SherpaOnnxOfflineSpeakerDiarizationResultGetNumSpeakers

(

const SherpaOnnxOfflineSpeakerDiarizationResult *

r

)

Parameters

r	A pointer returned by one of the SherpaOnnxOfflineSpeakerDiarizationProcess*() functions.

Returns

Number of speaker clusters.

SherpaOnnxOfflineSpeakerDiarizationResultSortByStartTime()

const SherpaOnnxOfflineSpeakerDiarizationSegment * SherpaOnnxOfflineSpeakerDiarizationResultSortByStartTime

(

const SherpaOnnxOfflineSpeakerDiarizationResult *

r

)

The returned array contains exactly SherpaOnnxOfflineSpeakerDiarizationResultGetNumSegments() entries.

Parameters

r	A pointer returned by one of the SherpaOnnxOfflineSpeakerDiarizationProcess*() functions.

Returns

A newly allocated segment array. Free it with SherpaOnnxOfflineSpeakerDiarizationDestroySegment().

SherpaOnnxOfflineSpeakerDiarizationSetConfig()

void SherpaOnnxOfflineSpeakerDiarizationSetConfig	(	const SherpaOnnxOfflineSpeakerDiarization *	sd,
		const SherpaOnnxOfflineSpeakerDiarizationConfig *	config
	)

Only config->clustering is used. Other fields are ignored.

Parameters

sd	A pointer returned by SherpaOnnxCreateOfflineSpeakerDiarization().
config	Configuration whose clustering field will be applied.

SherpaOnnxOfflineSpeechDenoiserGetSampleRate()

int32_t SherpaOnnxOfflineSpeechDenoiserGetSampleRate

(

const SherpaOnnxOfflineSpeechDenoiser *

sd

)

Parameters

sd	A pointer returned by SherpaOnnxCreateOfflineSpeechDenoiser().

Returns

Required input sample rate in Hz.

SherpaOnnxOfflineSpeechDenoiserRun()

const SherpaOnnxDenoisedAudio * SherpaOnnxOfflineSpeechDenoiserRun	(	const SherpaOnnxOfflineSpeechDenoiser *	sd,
		const float *	samples,
		int32_t	n,
		int32_t	sample_rate
	)

Parameters

sd	A pointer returned by SherpaOnnxCreateOfflineSpeechDenoiser().
samples	Input mono PCM samples normalized to [-1, 1].
n	Number of input samples.
sample_rate	Input sample rate in Hz.

Returns

A newly allocated denoised waveform. Free it with SherpaOnnxDestroyDenoisedAudio().

const SherpaOnnxDenoisedAudio *denoised =
    SherpaOnnxOfflineSpeechDenoiserRun(sd, wave->samples, wave->num_samples,
                                       wave->sample_rate);
SherpaOnnxWriteWave(denoised->samples, denoised->n, denoised->sample_rate,
                    "./enhanced.wav");
SherpaOnnxDestroyDenoisedAudio(denoised);

SherpaOnnxOfflineStreamGetOption()

const char * SherpaOnnxOfflineStreamGetOption	(	const SherpaOnnxOfflineStream *	stream,
		const char *	key
	)

Parameters

stream	A pointer returned by SherpaOnnxCreateOfflineStream().
key	Option name.

Returns

The option value. The returned pointer is owned by the stream, must not be freed by the caller, and may be invalidated if the option is overwritten or the stream is destroyed.

const char *value = SherpaOnnxOfflineStreamGetOption(stream, "language");

SherpaOnnxOfflineStreamHasOption()

int32_t SherpaOnnxOfflineStreamHasOption	(	const SherpaOnnxOfflineStream *	stream,
		const char *	key
	)

Parameters

stream	A pointer returned by SherpaOnnxCreateOfflineStream().
key	Option name.

Returns

1 if the option exists; otherwise 0.

int32_t has_language =
    SherpaOnnxOfflineStreamHasOption(stream, "language");

SherpaOnnxOfflineStreamSetOption()

void SherpaOnnxOfflineStreamSetOption	(	const SherpaOnnxOfflineStream *	stream,
		const char *	key,
		const char *	value
	)

Parameters

stream	A pointer returned by SherpaOnnxCreateOfflineStream().
key	Option name.
value	Option value represented as text.

SherpaOnnxOfflineStreamSetOption(stream, "language", "en");

SherpaOnnxOfflineTtsGenerate()

const SherpaOnnxGeneratedAudio * SherpaOnnxOfflineTtsGenerate	(	const SherpaOnnxOfflineTts *	tts,
		const char *	text,
		int32_t	sid,
		float	speed
	)

Deprecated:

Use SherpaOnnxOfflineTtsGenerateWithConfig() instead.

Parameters

tts	A pointer returned by SherpaOnnxCreateOfflineTts().
text	Input text.
sid	Speaker ID for multi-speaker models.
speed	Speech rate. Values > 1 are faster.

Returns

Generated audio, or NULL on error. Free it with SherpaOnnxDestroyOfflineTtsGeneratedAudio().

const SherpaOnnxGeneratedAudio *audio =
    SherpaOnnxOfflineTtsGenerate(tts, "Hello from sherpa-onnx!", 0, 1.0f);
SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                    "./generated.wav");
SherpaOnnxDestroyOfflineTtsGeneratedAudio(audio);

SherpaOnnxOfflineTtsGenerateWithCallback()

const SherpaOnnxGeneratedAudio * SherpaOnnxOfflineTtsGenerateWithCallback	(	const SherpaOnnxOfflineTts *	tts,
		const char *	text,
		int32_t	sid,
		float	speed,
		SherpaOnnxGeneratedAudioCallback	callback
	)

Deprecated:

Use SherpaOnnxOfflineTtsGenerateWithConfig() instead.

The callback receives newly generated samples. The sample pointer is valid only for the duration of the callback.

Parameters

tts	A pointer returned by SherpaOnnxCreateOfflineTts().
text	Input text.
sid	Speaker ID for multi-speaker models.
speed	Speech rate. Values > 1 are faster.
callback	Incremental callback. Return 0 to stop generation early.

Returns

Final generated audio, or NULL on error. Free it with SherpaOnnxDestroyOfflineTtsGeneratedAudio().

SherpaOnnxOfflineTtsGenerateWithCallbackWithArg()

const SherpaOnnxGeneratedAudio * SherpaOnnxOfflineTtsGenerateWithCallbackWithArg	(	const SherpaOnnxOfflineTts *	tts,
		const char *	text,
		int32_t	sid,
		float	speed,
		SherpaOnnxGeneratedAudioCallbackWithArg	callback,
		void *	arg
	)

Deprecated:

Use SherpaOnnxOfflineTtsGenerateWithConfig() instead.

Parameters

tts	A pointer returned by SherpaOnnxCreateOfflineTts().
text	Input text.
sid	Speaker ID for multi-speaker models.
speed	Speech rate. Values > 1 are faster.
callback	Incremental callback with user pointer.
arg	User pointer forwarded to callback.

Returns

Final generated audio, or NULL on error. Free it with SherpaOnnxDestroyOfflineTtsGeneratedAudio().

SherpaOnnxOfflineTtsGenerateWithConfig()

const SherpaOnnxGeneratedAudio * SherpaOnnxOfflineTtsGenerateWithConfig	(	const SherpaOnnxOfflineTts *	tts,
		const char *	text,
		const SherpaOnnxGenerationConfig *	config,
		SherpaOnnxGeneratedAudioProgressCallbackWithArg	callback,
		void *	arg
	)

This is the preferred API for new integrations. It supports callback-based progress reporting and model-specific options such as reference audio.

Parameters

tts	A pointer returned by SherpaOnnxCreateOfflineTts().
text	Input text.
config	Generation-time configuration.
callback	Optional progress callback with user pointer. Return 0 to stop early.
arg	User pointer forwarded to callback.

Returns

Generated audio, or NULL on error. Free it with SherpaOnnxDestroyOfflineTtsGeneratedAudio().

SherpaOnnxGenerationConfig cfg;
memset(&cfg, 0, sizeof(cfg));
cfg.sid = 0;
cfg.speed = 1.0f;
cfg.silence_scale = 0.2f;
 
const SherpaOnnxGeneratedAudio *audio =
    SherpaOnnxOfflineTtsGenerateWithConfig(tts,
        "Today as always, men fall into two groups.",
        &cfg, NULL, NULL);

SherpaOnnxOfflineTtsGenerateWithProgressCallback()

const SherpaOnnxGeneratedAudio * SherpaOnnxOfflineTtsGenerateWithProgressCallback	(	const SherpaOnnxOfflineTts *	tts,
		const char *	text,
		int32_t	sid,
		float	speed,
		SherpaOnnxGeneratedAudioProgressCallback	callback
	)

Deprecated:

Use SherpaOnnxOfflineTtsGenerateWithConfig() instead.

Parameters

tts	A pointer returned by SherpaOnnxCreateOfflineTts().
text	Input text.
sid	Speaker ID for multi-speaker models.
speed	Speech rate. Values > 1 are faster.
callback	Progress callback. Return 0 to stop generation early.

Returns

Final generated audio, or NULL on error. Free it with SherpaOnnxDestroyOfflineTtsGeneratedAudio().

int32_t Progress(const float *samples, int32_t n, float p) {
  fprintf(stderr, "Progress: %.2f%%\n", p * 100);
  return 1;
}
 
const SherpaOnnxGeneratedAudio *audio =
    SherpaOnnxOfflineTtsGenerateWithProgressCallback(tts, text, 0, 1.0f,
                                                     Progress);

SherpaOnnxOfflineTtsGenerateWithProgressCallbackWithArg()

const SherpaOnnxGeneratedAudio * SherpaOnnxOfflineTtsGenerateWithProgressCallbackWithArg	(	const SherpaOnnxOfflineTts *	tts,
		const char *	text,
		int32_t	sid,
		float	speed,
		SherpaOnnxGeneratedAudioProgressCallbackWithArg	callback,
		void *	arg
	)

Deprecated:

Use SherpaOnnxOfflineTtsGenerateWithConfig() instead.

Parameters

tts	A pointer returned by SherpaOnnxCreateOfflineTts().
text	Input text.
sid	Speaker ID for multi-speaker models.
speed	Speech rate. Values > 1 are faster.
callback	Progress callback with user pointer. Return 0 to stop early.
arg	User pointer forwarded to callback.

Returns

Final generated audio, or NULL on error. Free it with SherpaOnnxDestroyOfflineTtsGeneratedAudio().

SherpaOnnxOfflineTtsGenerateWithZipvoice()

const SherpaOnnxGeneratedAudio * SherpaOnnxOfflineTtsGenerateWithZipvoice	(	const SherpaOnnxOfflineTts *	tts,
		const char *	text,
		const char *	prompt_text,
		const float *	prompt_samples,
		int32_t	n_prompt,
		int32_t	prompt_sr,
		float	speed,
		int32_t	num_steps
	)

Use SherpaOnnxOfflineTtsGenerateWithConfig() instead.

SherpaOnnxOfflineTtsNumSpeakers()

int32_t SherpaOnnxOfflineTtsNumSpeakers

(

const SherpaOnnxOfflineTts *

tts

)

Single-speaker models often return 1.

Parameters

tts	A pointer returned by SherpaOnnxCreateOfflineTts().

Returns

Number of speakers supported by the model.

SherpaOnnxOfflineTtsSampleRate()

int32_t SherpaOnnxOfflineTtsSampleRate

(

const SherpaOnnxOfflineTts *

tts

)

Parameters

tts	A pointer returned by SherpaOnnxCreateOfflineTts().

Returns

Output sample rate in Hz.

SherpaOnnxOnlinePunctuationAddPunct()

const char * SherpaOnnxOnlinePunctuationAddPunct	(	const SherpaOnnxOnlinePunctuation *	punctuation,
		const char *	text
	)

Parameters

punctuation	A pointer returned by SherpaOnnxCreateOnlinePunctuation().
text	Input text chunk.

Returns

A newly allocated punctuated string. Free it with SherpaOnnxOnlinePunctuationFreeText().

const char *out =
    SherpaOnnxOnlinePunctuationAddPunct(punct,
        "how are you i am fine thank you");
printf("%s\n", out);
SherpaOnnxOnlinePunctuationFreeText(out);

SherpaOnnxOnlinePunctuationFreeText()

void SherpaOnnxOnlinePunctuationFreeText

(

const char *

text

)

Parameters

text	A pointer returned by SherpaOnnxOnlinePunctuationAddPunct().

SherpaOnnxOnlineSpeechDenoiserFlush()

const SherpaOnnxDenoisedAudio * SherpaOnnxOnlineSpeechDenoiserFlush

(

const SherpaOnnxOnlineSpeechDenoiser *

sd

)

This also resets the denoiser so it can be reused for a new utterance.

Parameters

sd	A pointer returned by SherpaOnnxCreateOnlineSpeechDenoiser().

Returns

A newly allocated denoised chunk, or NULL if no buffered output remains. Free non-NULL results with SherpaOnnxDestroyDenoisedAudio().

SherpaOnnxOnlineSpeechDenoiserGetFrameShiftInSamples()

int32_t SherpaOnnxOnlineSpeechDenoiserGetFrameShiftInSamples

(

const SherpaOnnxOnlineSpeechDenoiser *

sd

)

Example programs feed audio to the online denoiser in this chunk size.

Parameters

sd	A pointer returned by SherpaOnnxCreateOnlineSpeechDenoiser().

Returns

Frame shift in samples.

SherpaOnnxOnlineSpeechDenoiserGetSampleRate()

int32_t SherpaOnnxOnlineSpeechDenoiserGetSampleRate

(

const SherpaOnnxOnlineSpeechDenoiser *

sd

)

Parameters

sd	A pointer returned by SherpaOnnxCreateOnlineSpeechDenoiser().

Returns

Required input sample rate in Hz.

SherpaOnnxOnlineSpeechDenoiserReset()

void SherpaOnnxOnlineSpeechDenoiserReset

(

const SherpaOnnxOnlineSpeechDenoiser *

sd

)

Parameters

sd	A pointer returned by SherpaOnnxCreateOnlineSpeechDenoiser().

SherpaOnnxOnlineSpeechDenoiserRun()

const SherpaOnnxDenoisedAudio * SherpaOnnxOnlineSpeechDenoiserRun	(	const SherpaOnnxOnlineSpeechDenoiser *	sd,
		const float *	samples,
		int32_t	n,
		int32_t	sample_rate
	)

This function is not thread-safe. It may return NULL when not enough input has been accumulated to produce denoised output yet.

Parameters

sd	A pointer returned by SherpaOnnxCreateOnlineSpeechDenoiser().
samples	Input chunk normalized to [-1, 1].
n	Number of input samples.
sample_rate	Input sample rate in Hz.

Returns

A newly allocated denoised chunk, or NULL if no output is available yet. Free non-NULL results with SherpaOnnxDestroyDenoisedAudio().

SherpaOnnxOnlineStreamAcceptWaveform()

void SherpaOnnxOnlineStreamAcceptWaveform	(	const SherpaOnnxOnlineStream *	stream,
		int32_t	sample_rate,
		const float *	samples,
		int32_t	n
	)

The input is mono floating-point PCM normalized to the range [-1, 1]. If sample_rate differs from the recognizer feature sample rate, sherpa-onnx resamples internally.

Parameters

stream	A pointer returned by SherpaOnnxCreateOnlineStream().
sample_rate	Sample rate of samples.
samples	Pointer to n samples in the range [-1, 1].
n	Number of samples.

int32_t start = 0;
int32_t chunk_size = 3200;  // 0.2 seconds at 16 kHz
SherpaOnnxOnlineStreamAcceptWaveform(stream, wave->sample_rate,
                                     wave->samples + start, chunk_size);

SherpaOnnxOnlineStreamGetOption()

const char * SherpaOnnxOnlineStreamGetOption	(	const SherpaOnnxOnlineStream *	stream,
		const char *	key
	)

Parameters

stream	A pointer returned by SherpaOnnxCreateOnlineStream().
key	Option name.

Returns

The option value. The returned pointer is owned by the stream, must not be freed by the caller, and may be invalidated if the option is overwritten or the stream is destroyed.

const char *value = SherpaOnnxOnlineStreamGetOption(stream, "is_final");

SherpaOnnxOnlineStreamHasOption()

int32_t SherpaOnnxOnlineStreamHasOption	(	const SherpaOnnxOnlineStream *	stream,
		const char *	key
	)

Parameters

stream	A pointer returned by SherpaOnnxCreateOnlineStream().
key	Option name.

Returns

1 if the option exists; otherwise 0.

int32_t has_option = SherpaOnnxOnlineStreamHasOption(stream, "is_final");

SherpaOnnxOnlineStreamInputFinished()

void SherpaOnnxOnlineStreamInputFinished

(

const SherpaOnnxOnlineStream *

stream

)

After calling this function, do not append more samples to the stream.

Parameters

stream

A pointer returned by SherpaOnnxCreateOnlineStream().

SherpaOnnxOnlineStreamInputFinished(stream);

SherpaOnnxOnlineStreamIsEndpoint()

int32_t SherpaOnnxOnlineStreamIsEndpoint	(	const SherpaOnnxOnlineRecognizer *	recognizer,
		const SherpaOnnxOnlineStream *	stream
	)

Parameters

recognizer	A pointer returned by SherpaOnnxCreateOnlineRecognizer().
stream	A pointer returned by SherpaOnnxCreateOnlineStream().

Returns

1 if an endpoint is detected; otherwise 0.

if (SherpaOnnxOnlineStreamIsEndpoint(recognizer, stream)) {
  SherpaOnnxOnlineStreamReset(recognizer, stream);
}

SherpaOnnxOnlineStreamReset()

void SherpaOnnxOnlineStreamReset	(	const SherpaOnnxOnlineRecognizer *	recognizer,
		const SherpaOnnxOnlineStream *	stream
	)

This clears the decoder state for the stream so that it can be reused for a new utterance.

Parameters

recognizer	A pointer returned by SherpaOnnxCreateOnlineRecognizer().
stream	A pointer returned by SherpaOnnxCreateOnlineStream().

if (SherpaOnnxOnlineStreamIsEndpoint(recognizer, stream)) {
  SherpaOnnxOnlineStreamReset(recognizer, stream);
}

SherpaOnnxOnlineStreamSetOption()

void SherpaOnnxOnlineStreamSetOption	(	const SherpaOnnxOnlineStream *	stream,
		const char *	key,
		const char *	value
	)

This is a generic extension point for model-specific or runtime-specific options such as "is_final" for streaming Paraformer.

Parameters

stream	A pointer returned by SherpaOnnxCreateOnlineStream().
key	Option name.
value	Option value represented as text.

SherpaOnnxOnlineStreamSetOption(stream, "is_final", "1");

SherpaOnnxPrint()

void SherpaOnnxPrint	(	const SherpaOnnxDisplay *	display,
		int32_t	idx,
		const char *	s
	)

Parameters

display	A pointer returned by SherpaOnnxCreateDisplay().
idx	Segment or utterance index to print.
s	Text to print.

SherpaOnnxPrint(display, segment_id, r->text);

SherpaOnnxReadWave()

const SherpaOnnxWave * SherpaOnnxReadWave

(

const char *

filename

)

Parameters

filename

Input WAVE filename.

Returns

A newly allocated wave object, or NULL on error. Free it with SherpaOnnxFreeWave().

const SherpaOnnxWave *wave = SherpaOnnxReadWave("./Obama.wav");
if (wave) {
  printf("sample_rate=%d, num_samples=%d\n",
         wave->sample_rate, wave->num_samples);
  SherpaOnnxFreeWave(wave);
}

SherpaOnnxReadWaveFromBinaryData()

const SherpaOnnxWave * SherpaOnnxReadWaveFromBinaryData	(	const char *	data,
		int32_t	n
	)

Parameters

data	Pointer to the WAVE file bytes.
n	Size of data in bytes.

Returns

A newly allocated wave object, or NULL on error. Free it with SherpaOnnxFreeWave().

SherpaOnnxReadWaveMultiChannel()

const SherpaOnnxMultiChannelWave * SherpaOnnxReadWaveMultiChannel

(

const char *

filename

)

Parameters

filename

Input WAVE filename.

Returns

A newly allocated multi-channel wave object, or NULL on error. Free it with SherpaOnnxFreeMultiChannelWave().

SherpaOnnxResetKeywordStream()

void SherpaOnnxResetKeywordStream	(	const SherpaOnnxKeywordSpotter *	spotter,
		const SherpaOnnxOnlineStream *	stream
	)

The examples call this immediately after a successful trigger so the next keyword can be detected independently.

Parameters

spotter	A pointer returned by SherpaOnnxCreateKeywordSpotter().
stream	A pointer returned by SherpaOnnxCreateKeywordStream() or SherpaOnnxCreateKeywordStreamWithKeywords().

SherpaOnnxSpeakerEmbeddingExtractorComputeEmbedding()

const float * SherpaOnnxSpeakerEmbeddingExtractorComputeEmbedding	(	const SherpaOnnxSpeakerEmbeddingExtractor *	p,
		const SherpaOnnxOnlineStream *	s
	)

The returned vector has SherpaOnnxSpeakerEmbeddingExtractorDim(p) elements. Free it with SherpaOnnxSpeakerEmbeddingExtractorDestroyEmbedding().

Parameters

p	A pointer returned by SherpaOnnxCreateSpeakerEmbeddingExtractor().
s	A pointer returned by SherpaOnnxSpeakerEmbeddingExtractorCreateStream().

Returns

A newly allocated embedding vector.

const SherpaOnnxOnlineStream *stream =
    SherpaOnnxSpeakerEmbeddingExtractorCreateStream(ex);
SherpaOnnxOnlineStreamAcceptWaveform(stream, wave->sample_rate,
wave->samples, wave->num_samples);
SherpaOnnxOnlineStreamInputFinished(stream);
if (SherpaOnnxSpeakerEmbeddingExtractorIsReady(ex, stream)) {
  const float *v =
      SherpaOnnxSpeakerEmbeddingExtractorComputeEmbedding(ex, stream);
  SherpaOnnxSpeakerEmbeddingExtractorDestroyEmbedding(v);
}
SherpaOnnxDestroyOnlineStream(stream);

SherpaOnnxSpeakerEmbeddingExtractorCreateStream()

const SherpaOnnxOnlineStream * SherpaOnnxSpeakerEmbeddingExtractorCreateStream

(

const SherpaOnnxSpeakerEmbeddingExtractor *

p

)

Feed samples with SherpaOnnxOnlineStreamAcceptWaveform(), then call SherpaOnnxSpeakerEmbeddingExtractorIsReady() and SherpaOnnxSpeakerEmbeddingExtractorComputeEmbedding().

Parameters

p	A pointer returned by SherpaOnnxCreateSpeakerEmbeddingExtractor().

Returns

A newly allocated online stream. Free it with SherpaOnnxDestroyOnlineStream().

SherpaOnnxSpeakerEmbeddingExtractorDestroyEmbedding()

void SherpaOnnxSpeakerEmbeddingExtractorDestroyEmbedding

(

const float *

v

)

Parameters

v	A pointer returned by SherpaOnnxSpeakerEmbeddingExtractorComputeEmbedding().

SherpaOnnxSpeakerEmbeddingExtractorDim()

int32_t SherpaOnnxSpeakerEmbeddingExtractorDim

(

const SherpaOnnxSpeakerEmbeddingExtractor *

p

)

Parameters

p	A pointer returned by SherpaOnnxCreateSpeakerEmbeddingExtractor().

Returns

Embedding dimension.

SherpaOnnxSpeakerEmbeddingExtractorIsReady()

int32_t SherpaOnnxSpeakerEmbeddingExtractorIsReady	(	const SherpaOnnxSpeakerEmbeddingExtractor *	p,
		const SherpaOnnxOnlineStream *	s
	)

Parameters

p	A pointer returned by SherpaOnnxCreateSpeakerEmbeddingExtractor().
s	A pointer returned by SherpaOnnxSpeakerEmbeddingExtractorCreateStream().

Returns

1 if the stream is ready; otherwise 0.

SherpaOnnxSpeakerEmbeddingManagerAdd()

int32_t SherpaOnnxSpeakerEmbeddingManagerAdd	(	const SherpaOnnxSpeakerEmbeddingManager *	p,
		const char *	name,
		const float *	v
	)

Parameters

p	A pointer returned by SherpaOnnxCreateSpeakerEmbeddingManager().
name	Speaker name.
v	Embedding vector with exactly dim elements.

Returns

1 on success; 0 on error.

SherpaOnnxSpeakerEmbeddingManagerAddList()

int32_t SherpaOnnxSpeakerEmbeddingManagerAddList	(	const SherpaOnnxSpeakerEmbeddingManager *	p,
		const char *	name,
		const float **	v
	)

v is a NULL-terminated array of embedding pointers: v[0], v[1], ..., v[n - 1], followed by v[n] == NULL.

Parameters

p	A pointer returned by SherpaOnnxCreateSpeakerEmbeddingManager().
name	Speaker name.
v	NULL-terminated array of embedding pointers.

Returns

1 on success; 0 on error.

const float *spk1_vec[4] = {e1, e2, e3, NULL};
SherpaOnnxSpeakerEmbeddingManagerAddList(manager, "fangjun", spk1_vec);

SherpaOnnxSpeakerEmbeddingManagerAddListFlattened()

int32_t SherpaOnnxSpeakerEmbeddingManagerAddListFlattened	(	const SherpaOnnxSpeakerEmbeddingManager *	p,
		const char *	name,
		const float *	v,
		int32_t	n
	)

The input contains n embeddings laid out consecutively, so the total array length must be n * dim.

Parameters

p	A pointer returned by SherpaOnnxCreateSpeakerEmbeddingManager().
name	Speaker name.
v	Flattened embedding array.
n	Number of embeddings in v.

Returns

1 on success; 0 on error.

SherpaOnnxSpeakerEmbeddingManagerContains()

int32_t SherpaOnnxSpeakerEmbeddingManagerContains	(	const SherpaOnnxSpeakerEmbeddingManager *	p,
		const char *	name
	)

Parameters

p	A pointer returned by SherpaOnnxCreateSpeakerEmbeddingManager().
name	Speaker name.

Returns

1 if the speaker exists; otherwise 0.

SherpaOnnxSpeakerEmbeddingManagerFreeAllSpeakers()

void SherpaOnnxSpeakerEmbeddingManagerFreeAllSpeakers

(

const char *const *

names

)

Parameters

names

A pointer returned by SherpaOnnxSpeakerEmbeddingManagerGetAllSpeakers().

SherpaOnnxSpeakerEmbeddingManagerFreeBestMatches()

void SherpaOnnxSpeakerEmbeddingManagerFreeBestMatches

(

const SherpaOnnxSpeakerEmbeddingManagerBestMatchesResult *

r

)

Parameters

r	A pointer returned by SherpaOnnxSpeakerEmbeddingManagerGetBestMatches().

SherpaOnnxSpeakerEmbeddingManagerFreeSearch()

void SherpaOnnxSpeakerEmbeddingManagerFreeSearch

(

const char *

name

)

Parameters

name	A pointer returned by SherpaOnnxSpeakerEmbeddingManagerSearch().

SherpaOnnxSpeakerEmbeddingManagerGetAllSpeakers()

const char *const * SherpaOnnxSpeakerEmbeddingManagerGetAllSpeakers

(

const SherpaOnnxSpeakerEmbeddingManager *

p

)

The returned array is NULL-terminated. If no speakers are enrolled, the returned array still exists and its first element is NULL.

Parameters

p	A pointer returned by SherpaOnnxCreateSpeakerEmbeddingManager().

Returns

A newly allocated NULL-terminated array of speaker names. Free it with SherpaOnnxSpeakerEmbeddingManagerFreeAllSpeakers().

SherpaOnnxSpeakerEmbeddingManagerGetBestMatches()

const SherpaOnnxSpeakerEmbeddingManagerBestMatchesResult * SherpaOnnxSpeakerEmbeddingManagerGetBestMatches	(	const SherpaOnnxSpeakerEmbeddingManager *	p,
		const float *	v,
		float	threshold,
		int32_t	n
	)

Parameters

p	A pointer returned by SherpaOnnxCreateSpeakerEmbeddingManager().
v	Query embedding vector.
threshold	Minimum similarity threshold in the range [0, 1].
n	Maximum number of matches to return.

Returns

A newly allocated result object, or NULL if no matches are found. Free it with SherpaOnnxSpeakerEmbeddingManagerFreeBestMatches().

SherpaOnnxSpeakerEmbeddingManagerNumSpeakers()

int32_t SherpaOnnxSpeakerEmbeddingManagerNumSpeakers

(

const SherpaOnnxSpeakerEmbeddingManager *

p

)

Parameters

p	A pointer returned by SherpaOnnxCreateSpeakerEmbeddingManager().

Returns

Number of enrolled speakers.

SherpaOnnxSpeakerEmbeddingManagerRemove()

int32_t SherpaOnnxSpeakerEmbeddingManagerRemove	(	const SherpaOnnxSpeakerEmbeddingManager *	p,
		const char *	name
	)

Parameters

p	A pointer returned by SherpaOnnxCreateSpeakerEmbeddingManager().
name	Speaker name to remove.

Returns

1 if removed; otherwise 0. Returns 0 if the speaker does not exist.

SherpaOnnxSpeakerEmbeddingManagerSearch()

const char * SherpaOnnxSpeakerEmbeddingManagerSearch	(	const SherpaOnnxSpeakerEmbeddingManager *	p,
		const float *	v,
		float	threshold
	)

Parameters

p	A pointer returned by SherpaOnnxCreateSpeakerEmbeddingManager().
v	Query embedding vector.
threshold	Minimum similarity threshold in the range [0, 1].

Returns

A newly allocated speaker name on match, or NULL if no speaker passes the threshold. Free the returned name with SherpaOnnxSpeakerEmbeddingManagerFreeSearch().

SherpaOnnxSpeakerEmbeddingManagerVerify()

int32_t SherpaOnnxSpeakerEmbeddingManagerVerify	(	const SherpaOnnxSpeakerEmbeddingManager *	p,
		const char *	name,
		const float *	v,
		float	threshold
	)

Parameters

p	A pointer returned by SherpaOnnxCreateSpeakerEmbeddingManager().
name	Speaker name to compare against.
v	Query embedding vector.
threshold	Minimum similarity threshold in the range [0, 1].

Returns

1 if the speaker matches; otherwise 0.

SherpaOnnxSpokenLanguageIdentificationCompute()

const SherpaOnnxSpokenLanguageIdentificationResult * SherpaOnnxSpokenLanguageIdentificationCompute	(	const SherpaOnnxSpokenLanguageIdentification *	slid,
		const SherpaOnnxOfflineStream *	s
	)

Example:

SherpaOnnxOfflineStream *stream =
    SherpaOnnxSpokenLanguageIdentificationCreateOfflineStream(slid);
SherpaOnnxAcceptWaveformOffline(stream, wave->sample_rate, wave->samples,
                                wave->num_samples);
const SherpaOnnxSpokenLanguageIdentificationResult *result =
    SherpaOnnxSpokenLanguageIdentificationCompute(slid, stream);
printf("lang=%s\n", result->lang);
SherpaOnnxDestroySpokenLanguageIdentificationResult(result);
SherpaOnnxDestroyOfflineStream(stream);

Parameters

slid	A pointer returned by SherpaOnnxCreateSpokenLanguageIdentification().
s	A pointer returned by SherpaOnnxSpokenLanguageIdentificationCreateOfflineStream().

Returns

A newly allocated result object. Free it with SherpaOnnxDestroySpokenLanguageIdentificationResult().

SherpaOnnxSpokenLanguageIdentificationCreateOfflineStream()

SherpaOnnxOfflineStream * SherpaOnnxSpokenLanguageIdentificationCreateOfflineStream

(

const SherpaOnnxSpokenLanguageIdentification *

slid

)

Feed audio to the returned stream with SherpaOnnxAcceptWaveformOffline(), and then call SherpaOnnxSpokenLanguageIdentificationCompute().

Parameters

slid	A pointer returned by SherpaOnnxCreateSpokenLanguageIdentification().

Returns

A newly allocated offline stream. Free it with SherpaOnnxDestroyOfflineStream().

SherpaOnnxVoiceActivityDetectorAcceptWaveform()

void SherpaOnnxVoiceActivityDetectorAcceptWaveform	(	const SherpaOnnxVoiceActivityDetector *	p,
		const float *	samples,
		int32_t	n
	)

Input samples are mono floating-point PCM in the range [-1, 1].

Parameters

p	A pointer returned by SherpaOnnxCreateVoiceActivityDetector().
samples	Pointer to n samples.
n	Number of samples.

SherpaOnnxVoiceActivityDetectorAcceptWaveform(vad,
                                              wave->samples + i,
                                              window_size);

SherpaOnnxVoiceActivityDetectorClear()

void SherpaOnnxVoiceActivityDetectorClear

(

const SherpaOnnxVoiceActivityDetector *

p

)

Parameters

p	A pointer returned by SherpaOnnxCreateVoiceActivityDetector().

SherpaOnnxVoiceActivityDetectorDetected()

int32_t SherpaOnnxVoiceActivityDetectorDetected

(

const SherpaOnnxVoiceActivityDetector *

p

)

Parameters

p	A pointer returned by SherpaOnnxCreateVoiceActivityDetector().

Returns

1 if speech is currently detected; otherwise 0.

SherpaOnnxVoiceActivityDetectorEmpty()

int32_t SherpaOnnxVoiceActivityDetectorEmpty

(

const SherpaOnnxVoiceActivityDetector *

p

)

Parameters

p	A pointer returned by SherpaOnnxCreateVoiceActivityDetector().

Returns

1 if no completed speech segment is available; otherwise 0.

SherpaOnnxVoiceActivityDetectorFlush()

void SherpaOnnxVoiceActivityDetectorFlush

(

const SherpaOnnxVoiceActivityDetector *

p

)

Call this after the last chunk of input has been fed.

Parameters

p	A pointer returned by SherpaOnnxCreateVoiceActivityDetector().

SherpaOnnxVoiceActivityDetectorFlush(vad);

SherpaOnnxVoiceActivityDetectorFront()

const SherpaOnnxSpeechSegment * SherpaOnnxVoiceActivityDetectorFront

(

const SherpaOnnxVoiceActivityDetector *

p

)

The returned segment is a copy owned by the caller. Free it with SherpaOnnxDestroySpeechSegment().

Parameters

p	A pointer returned by SherpaOnnxCreateVoiceActivityDetector().

Returns

The first queued speech segment, or NULL if none is available.

while (!SherpaOnnxVoiceActivityDetectorEmpty(vad)) {
  const SherpaOnnxSpeechSegment *segment =
      SherpaOnnxVoiceActivityDetectorFront(vad);
  printf("start=%d, samples=%d\n", segment->start, segment->n);
  SherpaOnnxDestroySpeechSegment(segment);
  SherpaOnnxVoiceActivityDetectorPop(vad);
}

SherpaOnnxVoiceActivityDetectorPop()

void SherpaOnnxVoiceActivityDetectorPop

(

const SherpaOnnxVoiceActivityDetector *

p

)

Call this after consuming the segment returned by SherpaOnnxVoiceActivityDetectorFront().

Parameters

p	A pointer returned by SherpaOnnxCreateVoiceActivityDetector().

const SherpaOnnxSpeechSegment *segment =
    SherpaOnnxVoiceActivityDetectorFront(vad);
// ... use segment ...
SherpaOnnxDestroySpeechSegment(segment);
SherpaOnnxVoiceActivityDetectorPop(vad);

SherpaOnnxVoiceActivityDetectorReset()

void SherpaOnnxVoiceActivityDetectorReset

(

const SherpaOnnxVoiceActivityDetector *

p

)

Parameters

p	A pointer returned by SherpaOnnxCreateVoiceActivityDetector().

SherpaOnnxWaveFileSize()

int64_t SherpaOnnxWaveFileSize

(

int32_t

n_samples

)

Parameters

n_samples

Number of PCM samples.

Returns

Required buffer size in bytes.

SherpaOnnxWriteWave()

int32_t SherpaOnnxWriteWave	(	const float *	samples,
		int32_t	n,
		int32_t	sample_rate,
		const char *	filename
	)

Parameters

samples	Pointer to n samples in the range [-1, 1].
n	Number of samples.
sample_rate	Sample rate in Hz.
filename	Output filename.

Returns

1 on success; 0 on failure.

SherpaOnnxWriteWave(audio->samples, audio->n, audio->sample_rate,
                    "./generated-kokoro-en.wav");

SherpaOnnxWriteWaveMultiChannel()

int32_t SherpaOnnxWriteWaveMultiChannel	(	const float *const *	samples,
		int32_t	n,
		int32_t	sample_rate,
		int32_t	num_channels,
		const char *	filename
	)

Parameters

samples	samples[c] is a pointer to channel c samples in [-1, 1].
n	Number of samples per channel.
sample_rate	Sample rate in Hz.
num_channels	Number of channels.
filename	Output filename.

Returns

1 on success; 0 on failure.

SherpaOnnxWriteWaveToBuffer()

void SherpaOnnxWriteWaveToBuffer	(	const float *	samples,
		int32_t	n,
		int32_t	sample_rate,
		char *	buffer
	)

Allocate at least SherpaOnnxWaveFileSize(n) bytes before calling.

Parameters

samples	Pointer to n samples in the range [-1, 1].
n	Number of samples.
sample_rate	Sample rate in Hz.
buffer	Output buffer.