WebAssembly exmaple for speaker diarization (#1411)

This commit is contained in:
Fangjun Kuang
2024-10-10 22:14:45 +08:00
committed by GitHub
parent 67349b52f2
commit 1d061df355
37 changed files with 1008 additions and 24 deletions

View File

@@ -16,6 +16,11 @@ namespace SherpaOnnx
_handle = new HandleRef(this, h);
}
public void SetConfig(OfflineSpeakerDiarizationConfig config)
{
SherpaOnnxOfflineSpeakerDiarizationSetConfig(_handle.Handle, ref config);
}
public OfflineSpeakerDiarizationSegment[] Process(float[] samples)
{
IntPtr result = SherpaOnnxOfflineSpeakerDiarizationProcess(_handle.Handle, samples, samples.Length);
@@ -117,6 +122,9 @@ namespace SherpaOnnx
[DllImport(Dll.Filename)]
private static extern void SherpaOnnxOfflineSpeakerDiarizationDestroySegment(IntPtr handle);
[DllImport(Dll.Filename)]
private static extern void SherpaOnnxOfflineSpeakerDiarizationSetConfig(IntPtr handle, ref OfflineSpeakerDiarizationConfig config);
}
}

View File

@@ -1276,6 +1276,16 @@ func (sd *OfflineSpeakerDiarization) SampleRate() int {
return int(C.SherpaOnnxOfflineSpeakerDiarizationGetSampleRate(sd.impl))
}
// only config.Clustering is used. All other fields are ignored
func (sd *OfflineSpeakerDiarization) SetConfig(config *OfflineSpeakerDiarizationConfig) {
c := C.struct_SherpaOnnxOfflineSpeakerDiarizationConfig{}
c.clustering.num_clusters = C.int(config.Clustering.NumClusters)
c.clustering.threshold = C.float(config.Clustering.Threshold)
SherpaOnnxOfflineSpeakerDiarizationSetConfig(sd.impl, &c)
}
type OfflineSpeakerDiarizationSegment struct {
Start float32
End float32

View File

@@ -25,6 +25,11 @@ class OfflineSpeakerDiarization {
process(samples) {
return addon.offlineSpeakerDiarizationProcess(this.handle, samples);
}
setConfig(config) {
addon.offlineSpeakerDiarizationSetConfig(config);
this.config.clustering = config.clustering;
}
}
module.exports = {

View File

@@ -251,6 +251,46 @@ static Napi::Array OfflineSpeakerDiarizationProcessWrapper(
return ans;
}
static void OfflineSpeakerDiarizationSetConfigWrapper(
const Napi::CallbackInfo &info) {
Napi::Env env = info.Env();
if (info.Length() != 2) {
std::ostringstream os;
os << "Expect only 2 arguments. Given: " << info.Length();
Napi::TypeError::New(env, os.str()).ThrowAsJavaScriptException();
return;
}
if (!info[0].IsExternal()) {
Napi::TypeError::New(
env, "Argument 0 should be an offline speaker diarization pointer.")
.ThrowAsJavaScriptException();
return;
}
const SherpaOnnxOfflineSpeakerDiarization *sd =
info[0].As<Napi::External<SherpaOnnxOfflineSpeakerDiarization>>().Data();
if (!info[1].IsObject()) {
Napi::TypeError::New(env, "Expect an object as the argument")
.ThrowAsJavaScriptException();
return;
}
Napi::Object o = info[0].As<Napi::Object>();
SherpaOnnxOfflineSpeakerDiarizationConfig c;
memset(&c, 0, sizeof(c));
c.clustering = GetFastClusteringConfig(o);
SherpaOnnxOfflineSpeakerDiarizationSetConfig(sd, &c);
}
void InitNonStreamingSpeakerDiarization(Napi::Env env, Napi::Object exports) {
exports.Set(Napi::String::New(env, "createOfflineSpeakerDiarization"),
Napi::Function::New(env, CreateOfflineSpeakerDiarizationWrapper));
@@ -262,4 +302,8 @@ void InitNonStreamingSpeakerDiarization(Napi::Env env, Napi::Object exports) {
exports.Set(
Napi::String::New(env, "offlineSpeakerDiarizationProcess"),
Napi::Function::New(env, OfflineSpeakerDiarizationProcessWrapper));
exports.Set(
Napi::String::New(env, "offlineSpeakerDiarizationSetConfig"),
Napi::Function::New(env, OfflineSpeakerDiarizationSetConfigWrapper));
}