This repository has been archived on 2025-08-26. You can view files and clone it, but cannot push or open issues or pull requests.
Files
enginex_bi_series-sherpa-onnx/dotnet-examples/vad-non-streaming-asr-paraformer/Program.cs
东风破 00de2bd00b Refactor .Net example project (#1049)
Co-authored-by: 东风破 <birdfishs@163.com>
2024-06-24 10:10:13 +08:00

63 lines
2.2 KiB
C#

// Copyright (c) 2024 Xiaomi Corporation
//
// This file shows how to use a silero_vad model with a non-streaming Paraformer
// for speech recognition.
using SherpaOnnx;
using System.Collections.Generic;
using System;
class VadNonStreamingAsrParaformer
{
static void Main(string[] args)
{
// please download model files from
// https://github.com/k2-fsa/sherpa-onnx/releases/tag/asr-models
OfflineRecognizerConfig config = new OfflineRecognizerConfig();
config.ModelConfig.Paraformer.Model = "./sherpa-onnx-paraformer-zh-2023-03-28/model.int8.onnx";
config.ModelConfig.Tokens = "./sherpa-onnx-paraformer-zh-2023-03-28/tokens.txt";
config.ModelConfig.Debug = 0;
OfflineRecognizer recognizer = new OfflineRecognizer(config);
VadModelConfig vadModelConfig = new VadModelConfig();
vadModelConfig.SileroVad.Model = "./silero_vad.onnx";
vadModelConfig.Debug = 0;
VoiceActivityDetector vad = new VoiceActivityDetector(vadModelConfig, 60);
string testWaveFilename = "./lei-jun-test.wav";
WaveReader reader = new WaveReader(testWaveFilename);
int numSamples = reader.Samples.Length;
int windowSize = vadModelConfig.SileroVad.WindowSize;
int sampleRate = vadModelConfig.SampleRate;
int numIter = numSamples / windowSize;
for (int i = 0; i != numIter; ++i) {
int start = i * windowSize;
float[] samples = new float[windowSize];
Array.Copy(reader.Samples, start, samples, 0, windowSize);
vad.AcceptWaveform(samples);
if (vad.IsSpeechDetected()) {
while (!vad.IsEmpty()) {
SpeechSegment segment = vad.Front();
float startTime = segment.Start / (float)sampleRate;
float duration = segment.Samples.Length / (float)sampleRate;
OfflineStream stream = recognizer.CreateStream();
stream.AcceptWaveform(sampleRate, segment.Samples);
recognizer.Decode(stream);
String text = stream.Result.Text;
if (!String.IsNullOrEmpty(text)) {
Console.WriteLine("{0}--{1}: {2}", String.Format("{0:0.00}", startTime),
String.Format("{0:0.00}", startTime+duration), text);
}
vad.Pop();
}
}
}
}
}