EngineX-Iluvatar/enginex-mr_series-sherpa-onnx
Archived
16
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity

JavaScript API with WebAssembly for speaker diarization (#1414)

Browse Source 
#1408 uses [node-addon-api](https://github.com/nodejs/node-addon-api) to call C API from JavaScript, whereas this pull request uses WebAssembly to call C API from JavaScript.
This commit is contained in:
Fangjun Kuang
2024-10-11 11:40:10 +08:00
committed by GitHub
parent f1b311ee4f
commit eefc172095
9 changed files with 122 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
wasm/nodejs/CMakeLists.txt
View File

@@ -70,6 +70,17 @@ set(exported_functions
SherpaOnnxDestroySpeechSegment
SherpaOnnxVoiceActivityDetectorReset
SherpaOnnxVoiceActivityDetectorFlush
# Speaker diarization
SherpaOnnxCreateOfflineSpeakerDiarization
SherpaOnnxDestroyOfflineSpeakerDiarization
SherpaOnnxOfflineSpeakerDiarizationDestroyResult
SherpaOnnxOfflineSpeakerDiarizationDestroySegment
SherpaOnnxOfflineSpeakerDiarizationGetSampleRate
SherpaOnnxOfflineSpeakerDiarizationProcess
SherpaOnnxOfflineSpeakerDiarizationProcessWithCallback
SherpaOnnxOfflineSpeakerDiarizationResultGetNumSegments
SherpaOnnxOfflineSpeakerDiarizationResultSortByStartTime
SherpaOnnxOfflineSpeakerDiarizationSetConfig
#
SherpaOnnxFileExists
SherpaOnnxReadWave
@@ -109,6 +120,7 @@ install(
${CMAKE_SOURCE_DIR}/wasm/tts/sherpa-onnx-tts.js
${CMAKE_SOURCE_DIR}/wasm/kws/sherpa-onnx-kws.js
${CMAKE_SOURCE_DIR}/wasm/vad/sherpa-onnx-vad.js
${CMAKE_SOURCE_DIR}/wasm/speaker-diarization/sherpa-onnx-speaker-diarization.js
${CMAKE_SOURCE_DIR}/wasm/nodejs/sherpa-onnx-wave.js
"$<TARGET_FILE_DIR:sherpa-onnx-wasm-nodejs>/sherpa-onnx-wasm-nodejs.js"
"$<TARGET_FILE_DIR:sherpa-onnx-wasm-nodejs>/sherpa-onnx-wasm-nodejs.wasm"

4
wasm/speaker-diarization/assets/README.md
View File

@@ -12,7 +12,6 @@ Remember to rename the downloaded files.
The following is an example.
```bash
cd wasm/speaker-diarization/assets/
@@ -22,9 +21,6 @@ rm sherpa-onnx-pyannote-segmentation-3-0.tar.bz2
cp sherpa-onnx-pyannote-segmentation-3-0/model.onnx ./segmentation.onnx
rm -rf sherpa-onnx-pyannote-segmentation-3-0
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/speaker-recongition-models/3dspeaker_speech_eres2net_base_sv_zh-cn_3dspeaker_16k.onnx
mv 3dspeaker_speech_eres2net_base_sv_zh-cn_3dspeaker_16k.onnx ./embedding.onnx
```

12
wasm/speaker-diarization/sherpa-onnx-speaker-diarization.js
View File

@@ -64,7 +64,7 @@ function initSherpaOnnxOfflineSpeakerSegmentationModelConfig(config, Module) {
Module.setValue(ptr + offset, config.numThreads || 1, 'i32');
offset += 4;
Module.setValue(ptr + offset, config.debug || 1, 'i32');
Module.setValue(ptr + offset, config.debug || 0, 'i32');
offset += 4;
const providerLen = Module.lengthBytesUTF8(config.provider || 'cpu') + 1;
@@ -103,7 +103,7 @@ function initSherpaOnnxSpeakerEmbeddingExtractorConfig(config, Module) {
Module.setValue(ptr + offset, config.numThreads || 1, 'i32');
offset += 4;
Module.setValue(ptr + offset, config.debug || 1, 'i32');
Module.setValue(ptr + offset, config.debug || 0, 'i32');
offset += 4;
Module.setValue(ptr + offset, buffer + modelLen, 'i8*');
@@ -270,11 +270,15 @@ class OfflineSpeakerDiarization {
}
function createOfflineSpeakerDiarization(Module, myConfig) {
const config = {
let config = {
segmentation: {
pyannote: {model: './segmentation.onnx'},
debug: 1,
},
embedding: {
model: './embedding.onnx',
debug: 1,
},
embedding: {model: './embedding.onnx'},
clustering: {numClusters: -1, threshold: 0.5},
minDurationOn: 0.3,
minDurationOff: 0.5,