Add C# API for ten-vad (#2385)

.github/scripts/test-dot-net.sh

@@ -1,11 +1,20 @@
 #!/usr/bin/env bash
 
+set -ex
+
 cd dotnet-examples/
 
 cd ./version-test
 ./run.sh
 ls -lh
 
+cd ../vad-non-streaming-asr-paraformer
+./run-ten-vad.sh
+rm -fv *.onnx
+
+./run.sh
+rm -fv *.onnx
+
 cd ../non-streaming-canary-decode-files
 ./run.sh
 ls -lh
@@ -106,9 +115,6 @@ rm -rf sherpa-onnx-*
 ./run-paraformer.sh
 rm -rf sherpa-onnx-*
 
-cd ../vad-non-streaming-asr-paraformer
-./run.sh
-
 cd ../offline-punctuation
 ./run.sh
 rm -rf sherpa-onnx-*

.github/workflows/test-dot-net.yaml

@@ -102,6 +102,7 @@ jobs:
           df -h
 
       - name: Free space
+        if: false
         shell: bash
         run: |
           df -h
@@ -109,6 +110,7 @@ jobs:
           df -h
 
       - name: Free more space
+        if: false
         shell: bash
         run: |
           # https://github.com/orgs/community/discussions/25678
@@ -120,6 +122,7 @@ jobs:
           sudo rm -rf "$AGENT_TOOLSDIRECTORY"
 
       - name: Free Disk Space (Ubuntu)
+        if: false
         uses: jlumbroso/free-disk-space@main
         with:
           # this might remove tools that are actually needed,
@@ -136,6 +139,7 @@ jobs:
           swap-storage: true
 
       - name: Check space
+        if: false
         shell: bash
         run: |
           df -h

dotnet-examples/vad-non-streaming-asr-paraformer/Program.cs

@@ -1,8 +1,10 @@
 // Copyright (c)  2024  Xiaomi Corporation
 //
-// This file shows how to use a silero_vad model with a non-streaming Paraformer
-// for speech recognition.
+// This file shows how to use a silero_vad model or ten-vad model
+// with a non-streaming Paraformer for speech recognition.
 using SherpaOnnx;
+using System.IO;
+
 
 class VadNonStreamingAsrParaformer
 {
@@ -17,7 +19,31 @@ class VadNonStreamingAsrParaformer
     var recognizer = new OfflineRecognizer(config);
 
     var vadModelConfig = new VadModelConfig();
-    vadModelConfig.SileroVad.Model = "./silero_vad.onnx";
+    if (File.Exists("./silero_vad.onnx"))
+    {
+      Console.WriteLine("Use silero-vad");
+      vadModelConfig.SileroVad.Model = "./silero_vad.onnx";
+      vadModelConfig.SileroVad.Threshold = 0.3F;
+      vadModelConfig.SileroVad.MinSilenceDuration = 0.5F;
+      vadModelConfig.SileroVad.MinSpeechDuration = 0.25F;
+      vadModelConfig.SileroVad.MaxSpeechDuration = 5.0F;
+      vadModelConfig.SileroVad.WindowSize = 512;
+    }
+    else if (File.Exists("./ten-vad.onnx"))
+    {
+      Console.WriteLine("Use ten-vad");
+      vadModelConfig.TenVad.Model = "./ten-vad.onnx";
+      vadModelConfig.TenVad.Threshold = 0.3F;
+      vadModelConfig.TenVad.MinSilenceDuration = 0.5F;
+      vadModelConfig.TenVad.MinSpeechDuration = 0.25F;
+      vadModelConfig.TenVad.MaxSpeechDuration = 5.0F;
+      vadModelConfig.TenVad.WindowSize = 256;
+    }
+    else
+    {
+      Console.WriteLine("Please download ./silero_vad.onnx or ./ten-vad.onnx");
+      return;
+    }
     vadModelConfig.Debug = 0;
 
     var vad = new VoiceActivityDetector(vadModelConfig, 60);
@@ -27,6 +53,12 @@ class VadNonStreamingAsrParaformer
 
     int numSamples = reader.Samples.Length;
     int windowSize = vadModelConfig.SileroVad.WindowSize;
+
+    if (vadModelConfig.TenVad.Model != "")
+    {
+      windowSize = vadModelConfig.TenVad.WindowSize;
+    }
+
     int sampleRate = vadModelConfig.SampleRate;
     int numIter = numSamples / windowSize;
 

dotnet-examples/vad-non-streaming-asr-paraformer/run-ten-vad.sh

@@ -0,0 +1,20 @@
+#!/usr/bin/env bash
+
+set -ex
+
+if [ ! -f ./ten-vad.onnx ]; then
+  curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/ten-vad.onnx
+fi
+
+if [ ! -f ./lei-jun-test.wav ]; then
+  curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/lei-jun-test.wav
+fi
+
+if [ ! -f ./sherpa-onnx-paraformer-zh-2023-09-14/tokens.txt ]; then
+  curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-paraformer-zh-2023-09-14.tar.bz2
+
+  tar xvf sherpa-onnx-paraformer-zh-2023-09-14.tar.bz2
+  rm sherpa-onnx-paraformer-zh-2023-09-14.tar.bz2
+fi
+
+dotnet run

scripts/dotnet/TenVadModelConfig.cs

@@ -0,0 +1,33 @@
+/// Copyright (c)  2025  Xiaomi Corporation (authors: Fangjun Kuang)
+
+using System.Runtime.InteropServices;
+
+namespace SherpaOnnx
+{
+    [StructLayout(LayoutKind.Sequential)]
+    public struct TenVadModelConfig
+    {
+        public TenVadModelConfig()
+        {
+            Model = "";
+            Threshold = 0.5F;
+            MinSilenceDuration = 0.5F;
+            MinSpeechDuration = 0.25F;
+            WindowSize = 256;
+            MaxSpeechDuration = 5.0F;
+        }
+
+        [MarshalAs(UnmanagedType.LPStr)]
+        public string Model;
+
+        public float Threshold;
+
+        public float MinSilenceDuration;
+
+        public float MinSpeechDuration;
+
+        public int WindowSize;
+
+        public float MaxSpeechDuration;
+    }
+}

scripts/dotnet/VadModelConfig.cs

@@ -14,6 +14,7 @@ namespace SherpaOnnx
             NumThreads = 1;
             Provider = "cpu";
             Debug = 0;
+            TenVad = new TenVadModelConfig();
         }
 
         public SileroVadModelConfig SileroVad;
@@ -26,6 +27,8 @@ namespace SherpaOnnx
         public string Provider;
 
         public int Debug;
+
+        public TenVadModelConfig TenVad;
     }
 }