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

Refactor .Net example project (#1049)

Browse Source 
Co-authored-by: 东风破 <birdfishs@163.com>
This commit is contained in:
东风破
2024-06-24 10:10:13 +08:00
committed by GitHub
parent 6b7ced2317
commit 00de2bd00b
77 changed files with 526 additions and 651 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
dotnet-examples/.gitignore vendored
View File

@@ -1,3 +1,6 @@
bin
obj
v17
.vs
!*.sh
*.vsidx

13
dotnet-examples/Common/Common.csproj Normal file
View File

@@ -0,0 +1,13 @@
<Project Sdk="Microsoft.NET.Sdk">
<PropertyGroup>
<TargetFramework>net6.0</TargetFramework>
<AllowUnsafeBlocks>true</AllowUnsafeBlocks>
<RestoreSources>/tmp/packages;$(RestoreSources);https://api.nuget.org/v3/index.json</RestoreSources>
</PropertyGroup>
<ItemGroup>
<PackageReference Include="CommandLineParser" Version="2.9.1" />
<PackageReference Include="org.k2fsa.sherpa.onnx" Version="1.10.1" />
</ItemGroup>
</Project>

348
dotnet-examples/online-decode-files/WaveReader.cs → dotnet-examples/Common/WaveHeader.cs
View File

@@ -1,174 +1,174 @@
// Copyright (c) 2023 Xiaomi Corporation (authors: Fangjun Kuang)
using System;
using System.IO;
using System.Runtime.InteropServices;
namespace SherpaOnnx
{
[StructLayout(LayoutKind.Sequential)]
public struct WaveHeader
{
public Int32 ChunkID;
public Int32 ChunkSize;
public Int32 Format;
public Int32 SubChunk1ID;
public Int32 SubChunk1Size;
public Int16 AudioFormat;
public Int16 NumChannels;
public Int32 SampleRate;
public Int32 ByteRate;
public Int16 BlockAlign;
public Int16 BitsPerSample;
public Int32 SubChunk2ID;
public Int32 SubChunk2Size;
public bool Validate()
{
if (ChunkID != 0x46464952)
{
Console.WriteLine($"Invalid chunk ID: 0x{ChunkID:X}. Expect 0x46464952");
return false;
}
// E V A W
if (Format != 0x45564157)
{
Console.WriteLine($"Invalid format: 0x{Format:X}. Expect 0x45564157");
return false;
}
// t m f
if (SubChunk1ID != 0x20746d66)
{
Console.WriteLine($"Invalid SubChunk1ID: 0x{SubChunk1ID:X}. Expect 0x20746d66");
return false;
}
if (SubChunk1Size != 16)
{
Console.WriteLine($"Invalid SubChunk1Size: {SubChunk1Size}. Expect 16");
return false;
}
if (AudioFormat != 1)
{
Console.WriteLine($"Invalid AudioFormat: {AudioFormat}. Expect 1");
return false;
}
if (NumChannels != 1)
{
Console.WriteLine($"Invalid NumChannels: {NumChannels}. Expect 1");
return false;
}
if (ByteRate != (SampleRate * NumChannels * BitsPerSample / 8))
{
Console.WriteLine($"Invalid byte rate: {ByteRate}.");
return false;
}
if (BlockAlign != (NumChannels * BitsPerSample / 8))
{
Console.WriteLine($"Invalid block align: {ByteRate}.");
return false;
}
if (BitsPerSample != 16)
{ // we support only 16 bits per sample
Console.WriteLine($"Invalid bits per sample: {BitsPerSample}. Expect 16");
return false;
}
return true;
}
}
// It supports only 16-bit, single channel WAVE format.
// The sample rate can be any value.
public class WaveReader
{
public WaveReader(String fileName)
{
if (!File.Exists(fileName))
{
throw new ApplicationException($"{fileName} does not exist!");
}
using (var stream = File.Open(fileName, FileMode.Open))
{
using (var reader = new BinaryReader(stream))
{
_header = ReadHeader(reader);
if (!_header.Validate())
{
throw new ApplicationException($"Invalid wave file ${fileName}");
}
SkipMetaData(reader);
// now read samples
// _header.SubChunk2Size contains number of bytes in total.
// we assume each sample is of type int16
byte[] buffer = reader.ReadBytes(_header.SubChunk2Size);
short[] samples_int16 = new short[_header.SubChunk2Size / 2];
Buffer.BlockCopy(buffer, 0, samples_int16, 0, buffer.Length);
_samples = new float[samples_int16.Length];
for (var i = 0; i < samples_int16.Length; ++i)
{
_samples[i] = samples_int16[i] / 32768.0F;
}
}
}
}
private static WaveHeader ReadHeader(BinaryReader reader)
{
byte[] bytes = reader.ReadBytes(Marshal.SizeOf(typeof(WaveHeader)));
GCHandle handle = GCHandle.Alloc(bytes, GCHandleType.Pinned);
WaveHeader header = (WaveHeader)Marshal.PtrToStructure(handle.AddrOfPinnedObject(), typeof(WaveHeader))!;
handle.Free();
return header;
}
private void SkipMetaData(BinaryReader reader)
{
var bs = reader.BaseStream;
Int32 subChunk2ID = _header.SubChunk2ID;
Int32 subChunk2Size = _header.SubChunk2Size;
while (bs.Position != bs.Length && subChunk2ID != 0x61746164)
{
bs.Seek(subChunk2Size, SeekOrigin.Current);
subChunk2ID = reader.ReadInt32();
subChunk2Size = reader.ReadInt32();
}
_header.SubChunk2ID = subChunk2ID;
_header.SubChunk2Size = subChunk2Size;
}
private WaveHeader _header;
// Samples are normalized to the range [-1, 1]
private float[] _samples;
public int SampleRate => _header.SampleRate;
public float[] Samples => _samples;
public static void Test(String fileName)
{
WaveReader reader = new WaveReader(fileName);
Console.WriteLine($"samples length: {reader.Samples.Length}");
Console.WriteLine($"samples rate: {reader.SampleRate}");
}
}
}
// Copyright (c) 2023 Xiaomi Corporation (authors: Fangjun Kuang)
using System;
using System.IO;
using System.Runtime.InteropServices;
namespace SherpaOnnx
{
[StructLayout(LayoutKind.Sequential)]
public struct WaveHeader
{
public Int32 ChunkID;
public Int32 ChunkSize;
public Int32 Format;
public Int32 SubChunk1ID;
public Int32 SubChunk1Size;
public Int16 AudioFormat;
public Int16 NumChannels;
public Int32 SampleRate;
public Int32 ByteRate;
public Int16 BlockAlign;
public Int16 BitsPerSample;
public Int32 SubChunk2ID;
public Int32 SubChunk2Size;
public bool Validate()
{
if (ChunkID != 0x46464952)
{
Console.WriteLine($"Invalid chunk ID: 0x{ChunkID:X}. Expect 0x46464952");
return false;
}
// E V A W
if (Format != 0x45564157)
{
Console.WriteLine($"Invalid format: 0x{Format:X}. Expect 0x45564157");
return false;
}
// t m f
if (SubChunk1ID != 0x20746d66)
{
Console.WriteLine($"Invalid SubChunk1ID: 0x{SubChunk1ID:X}. Expect 0x20746d66");
return false;
}
if (SubChunk1Size != 16)
{
Console.WriteLine($"Invalid SubChunk1Size: {SubChunk1Size}. Expect 16");
return false;
}
if (AudioFormat != 1)
{
Console.WriteLine($"Invalid AudioFormat: {AudioFormat}. Expect 1");
return false;
}
if (NumChannels != 1)
{
Console.WriteLine($"Invalid NumChannels: {NumChannels}. Expect 1");
return false;
}
if (ByteRate != (SampleRate * NumChannels * BitsPerSample / 8))
{
Console.WriteLine($"Invalid byte rate: {ByteRate}.");
return false;
}
if (BlockAlign != (NumChannels * BitsPerSample / 8))
{
Console.WriteLine($"Invalid block align: {ByteRate}.");
return false;
}
if (BitsPerSample != 16)
{ // we support only 16 bits per sample
Console.WriteLine($"Invalid bits per sample: {BitsPerSample}. Expect 16");
return false;
}
return true;
}
}
// It supports only 16-bit, single channel WAVE format.
// The sample rate can be any value.
public class WaveReader
{
public WaveReader(String fileName)
{
if (!File.Exists(fileName))
{
throw new ApplicationException($"{fileName} does not exist!");
}
using (var stream = File.Open(fileName, FileMode.Open))
{
using (var reader = new BinaryReader(stream))
{
_header = ReadHeader(reader);
if (!_header.Validate())
{
throw new ApplicationException($"Invalid wave file ${fileName}");
}
SkipMetaData(reader);
// now read samples
// _header.SubChunk2Size contains number of bytes in total.
// we assume each sample is of type int16
byte[] buffer = reader.ReadBytes(_header.SubChunk2Size);
short[] samples_int16 = new short[_header.SubChunk2Size / 2];
Buffer.BlockCopy(buffer, 0, samples_int16, 0, buffer.Length);
_samples = new float[samples_int16.Length];
for (var i = 0; i < samples_int16.Length; ++i)
{
_samples[i] = samples_int16[i] / 32768.0F;
}
}
}
}
private static WaveHeader ReadHeader(BinaryReader reader)
{
byte[] bytes = reader.ReadBytes(Marshal.SizeOf(typeof(WaveHeader)));
GCHandle handle = GCHandle.Alloc(bytes, GCHandleType.Pinned);
WaveHeader header = (WaveHeader)Marshal.PtrToStructure(handle.AddrOfPinnedObject(), typeof(WaveHeader))!;
handle.Free();
return header;
}
private void SkipMetaData(BinaryReader reader)
{
var bs = reader.BaseStream;
Int32 subChunk2ID = _header.SubChunk2ID;
Int32 subChunk2Size = _header.SubChunk2Size;
while (bs.Position != bs.Length && subChunk2ID != 0x61746164)
{
bs.Seek(subChunk2Size, SeekOrigin.Current);
subChunk2ID = reader.ReadInt32();
subChunk2Size = reader.ReadInt32();
}
_header.SubChunk2ID = subChunk2ID;
_header.SubChunk2Size = subChunk2Size;
}
private WaveHeader _header;
// Samples are normalized to the range [-1, 1]
private float[] _samples;
public int SampleRate => _header.SampleRate;
public float[] Samples => _samples;
public static void Test(String fileName)
{
WaveReader reader = new WaveReader(fileName);
Console.WriteLine($"samples length: {reader.Samples.Length}");
Console.WriteLine($"samples rate: {reader.SampleRate}");
}
}
}

1
dotnet-examples/offline-decode-files/WaveReader.cs
View File

@@ -1 +0,0 @@
../online-decode-files/WaveReader.cs

3
dotnet-examples/offline-decode-files/offline-decode-files.csproj
View File

@@ -9,8 +9,7 @@
</PropertyGroup>
<ItemGroup>
<PackageReference Include="CommandLineParser" Version="2.9.1" />
<PackageReference Include="org.k2fsa.sherpa.onnx" Version="*" />
<ProjectReference Include="..\Common\Common.csproj" />
</ItemGroup>
</Project>

86
dotnet-examples/offline-punctuation/Program.cs
View File

@@ -1,43 +1,43 @@
// Copyright (c) 2024 Xiaomi Corporation
//
// This file shows how to add punctuations to text.
//
// 1. Download a model from
// https://github.com/k2-fsa/sherpa-onnx/releases/tag/punctuation-models
//
// wget https://github.com/k2-fsa/sherpa-onnx/releases/download/punctuation-models/sherpa-onnx-punct-ct-transformer-zh-en-vocab272727-2024-04-12.tar.bz2
//
// 3. Now run it
//
// dotnet run
using SherpaOnnx;
using System.Collections.Generic;
using System;
class OfflinePunctuationDemo
{
static void Main(string[] args)
{
var config = new OfflinePunctuationConfig();
config.Model.CtTransformer = "./sherpa-onnx-punct-ct-transformer-zh-en-vocab272727-2024-04-12/model.onnx";
config.Model.Debug = 1;
config.Model.NumThreads = 1;
var punct = new OfflinePunctuation(config);
string[] textList = new string[] {
"这是一个测试你好吗How are you我很好thank you are you ok谢谢你",
"我们都是木头人不会说话不会动",
"The African blogosphere is rapidly expanding bringing more voices online in the form of commentaries opinions analyses rants and poetry",
};
Console.WriteLine("---------");
foreach (string text in textList)
{
string textWithPunct = punct.AddPunct(text);
Console.WriteLine("Input text: {0}", text);
Console.WriteLine("Output text: {0}", textWithPunct);
Console.WriteLine("---------");
}
}
}
// Copyright (c) 2024 Xiaomi Corporation
//
// This file shows how to add punctuations to text.
//
// 1. Download a model from
// https://github.com/k2-fsa/sherpa-onnx/releases/tag/punctuation-models
//
// wget https://github.com/k2-fsa/sherpa-onnx/releases/download/punctuation-models/sherpa-onnx-punct-ct-transformer-zh-en-vocab272727-2024-04-12.tar.bz2
//
// 3. Now run it
//
// dotnet run
using SherpaOnnx;
using System.Collections.Generic;
using System;
class OfflinePunctuationDemo
{
static void Main(string[] args)
{
var config = new OfflinePunctuationConfig();
config.Model.CtTransformer = "./sherpa-onnx-punct-ct-transformer-zh-en-vocab272727-2024-04-12/model.onnx";
config.Model.Debug = 1;
config.Model.NumThreads = 1;
var punct = new OfflinePunctuation(config);
string[] textList = new string[] {
"这是一个测试你好吗How are you我很好thank you are you ok谢谢你",
"我们都是木头人不会说话不会动",
"The African blogosphere is rapidly expanding bringing more voices online in the form of commentaries opinions analyses rants and poetry",
};
Console.WriteLine("---------");
foreach (string text in textList)
{
string textWithPunct = punct.AddPunct(text);
Console.WriteLine("Input text: {0}", text);
Console.WriteLine("Output text: {0}", textWithPunct);
Console.WriteLine("---------");
}
}
}

30
dotnet-examples/offline-punctuation/offline-punctuation.csproj
View File

@@ -1,15 +1,15 @@
<Project Sdk="Microsoft.NET.Sdk">
<PropertyGroup>
<OutputType>Exe</OutputType>
<TargetFramework>net6.0</TargetFramework>
<RootNamespace>offline_punctuation</RootNamespace>
<ImplicitUsings>enable</ImplicitUsings>
<Nullable>enable</Nullable>
</PropertyGroup>
<ItemGroup>
<PackageReference Include="org.k2fsa.sherpa.onnx" Version="*" />
</ItemGroup>
</Project>
<Project Sdk="Microsoft.NET.Sdk">
<PropertyGroup>
<OutputType>Exe</OutputType>
<TargetFramework>net6.0</TargetFramework>
<RootNamespace>offline_punctuation</RootNamespace>
<ImplicitUsings>enable</ImplicitUsings>
<Nullable>enable</Nullable>
</PropertyGroup>
<ItemGroup>
<ProjectReference Include="..\Common\Common.csproj" />
</ItemGroup>
</Project>

0
dotnet-examples/offline-punctuation/run.sh Executable file → Normal file
View File

6
dotnet-examples/offline-tts-play/offline-tts-play.csproj
View File

@@ -9,9 +9,11 @@
</PropertyGroup>
<ItemGroup>
<PackageReference Include="CommandLineParser" Version="2.9.1" />
<PackageReference Include="org.k2fsa.sherpa.onnx" Version="*" />
<PackageReference Include="PortAudioSharp2" Version="*" />
</ItemGroup>
<ItemGroup>
<ProjectReference Include="..\Common\Common.csproj" />
</ItemGroup>
</Project>

3
dotnet-examples/offline-tts/offline-tts.csproj
View File

@@ -9,8 +9,7 @@
</PropertyGroup>
<ItemGroup>
<PackageReference Include="CommandLineParser" Version="2.9.1" />
<PackageReference Include="org.k2fsa.sherpa.onnx" Version="*" />
<ProjectReference Include="..\Common\Common.csproj" />
</ItemGroup>
</Project>

3
dotnet-examples/online-decode-files/online-decode-files.csproj
View File

@@ -9,8 +9,7 @@
</PropertyGroup>
<ItemGroup>
<PackageReference Include="CommandLineParser" Version="2.9.1" />
<PackageReference Include="org.k2fsa.sherpa.onnx" Version="*" />
<ProjectReference Include="..\Common\Common.csproj" />
</ItemGroup>
</Project>

35
dotnet-examples/sherpa-onnx.sln
View File

@@ -3,34 +3,33 @@ Microsoft Visual Studio Solution File, Format Version 12.00
# Visual Studio Version 17
VisualStudioVersion = 17.0.31903.59
MinimumVisualStudioVersion = 10.0.40219.1
Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "online-decode-files", "online-decode-files\online-decode-files.csproj", "{45307474-BECB-4ABE-9388-D01D55A1A9BE}"
Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "online-decode-files", "online-decode-files\online-decode-files.csproj", "{45307474-BECB-4ABE-9388-D01D55A1A9BE}"
EndProject
Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "offline-decode-files", "offline-decode-files\offline-decode-files.csproj", "{2DAB152C-9E24-47A0-9DB0-781297ECE458}"
Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "offline-decode-files", "offline-decode-files\offline-decode-files.csproj", "{2DAB152C-9E24-47A0-9DB0-781297ECE458}"
EndProject
Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "speech-recognition-from-microphone", "speech-recognition-from-microphone\speech-recognition-from-microphone.csproj", "{FE4EA1FF-062A-46B3-B78D-C828FED7B82E}"
Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "speech-recognition-from-microphone", "speech-recognition-from-microphone\speech-recognition-from-microphone.csproj", "{FE4EA1FF-062A-46B3-B78D-C828FED7B82E}"
EndProject
Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "offline-tts", "offline-tts\offline-tts.csproj", "{72196886-7143-4043-96E2-BCACEC6C79EB}"
Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "offline-tts", "offline-tts\offline-tts.csproj", "{72196886-7143-4043-96E2-BCACEC6C79EB}"
EndProject
Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "offline-tts-play", "offline-tts-play\offline-tts-play.csproj", "{40781464-5948-462B-BA4B-98932711513F}"
Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "offline-tts-play", "offline-tts-play\offline-tts-play.csproj", "{40781464-5948-462B-BA4B-98932711513F}"
EndProject
Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "spoken-language-identification", "spoken-language-identification\spoken-language-identification.csproj", "{3D7CF3D6-AC45-4D50-9619-5687B1443E94}"
Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "spoken-language-identification", "spoken-language-identification\spoken-language-identification.csproj", "{3D7CF3D6-AC45-4D50-9619-5687B1443E94}"
EndProject
Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "streaming-hlg-decoding", "streaming-hlg-decoding\streaming-hlg-decoding.csproj", "{C4A368A5-FCA0-419D-97C9-C8CE0B08EB99}"
Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "streaming-hlg-decoding", "streaming-hlg-decoding\streaming-hlg-decoding.csproj", "{C4A368A5-FCA0-419D-97C9-C8CE0B08EB99}"
EndProject
Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "speaker-identification", "speaker-identification\speaker-identification.csproj", "{2B1B140E-A92F-426B-B0DF-5D916B67304F}"
Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "speaker-identification", "speaker-identification\speaker-identification.csproj", "{2B1B140E-A92F-426B-B0DF-5D916B67304F}"
EndProject
Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "offline-punctuation", "offline-punctuation\offline-punctuation.csproj", "{42D85582-BB63-4259-A4EA-837D66AC078B}"
Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "offline-punctuation", "offline-punctuation\offline-punctuation.csproj", "{42D85582-BB63-4259-A4EA-837D66AC078B}"
EndProject
Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "vad-non-streaming-asr-paraformer", "vad-non-streaming-asr-paraformer\vad-non-streaming-asr-paraformer.csproj", "{8CD6B7E5-F59F-47B3-BB87-2B2E3678924D}"
Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "vad-non-streaming-asr-paraformer", "vad-non-streaming-asr-paraformer\vad-non-streaming-asr-paraformer.csproj", "{8CD6B7E5-F59F-47B3-BB87-2B2E3678924D}"
EndProject
Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "Common", "Common\Common.csproj", "{401E963F-E25A-43CE-987D-8DB2D4715756}"
EndProject
Global
GlobalSection(SolutionConfigurationPlatforms) = preSolution
Debug|Any CPU = Debug|Any CPU
Release|Any CPU = Release|Any CPU
EndGlobalSection
GlobalSection(SolutionProperties) = preSolution
HideSolutionNode = FALSE
EndGlobalSection
GlobalSection(ProjectConfigurationPlatforms) = postSolution
{45307474-BECB-4ABE-9388-D01D55A1A9BE}.Debug|Any CPU.ActiveCfg = Debug|Any CPU
{45307474-BECB-4ABE-9388-D01D55A1A9BE}.Debug|Any CPU.Build.0 = Debug|Any CPU
@@ -72,5 +71,15 @@ Global
{8CD6B7E5-F59F-47B3-BB87-2B2E3678924D}.Debug|Any CPU.Build.0 = Debug|Any CPU
{8CD6B7E5-F59F-47B3-BB87-2B2E3678924D}.Release|Any CPU.ActiveCfg = Release|Any CPU
{8CD6B7E5-F59F-47B3-BB87-2B2E3678924D}.Release|Any CPU.Build.0 = Release|Any CPU
{401E963F-E25A-43CE-987D-8DB2D4715756}.Debug|Any CPU.ActiveCfg = Debug|Any CPU
{401E963F-E25A-43CE-987D-8DB2D4715756}.Debug|Any CPU.Build.0 = Debug|Any CPU
{401E963F-E25A-43CE-987D-8DB2D4715756}.Release|Any CPU.ActiveCfg = Release|Any CPU
{401E963F-E25A-43CE-987D-8DB2D4715756}.Release|Any CPU.Build.0 = Release|Any CPU
EndGlobalSection
GlobalSection(SolutionProperties) = preSolution
HideSolutionNode = FALSE
EndGlobalSection
GlobalSection(ExtensibilityGlobals) = postSolution
SolutionGuid = {07A6023C-0A37-4F82-A29F-896A3A338EAC}
EndGlobalSection
EndGlobal

1
dotnet-examples/speaker-identification/WaveReader.cs
View File

@@ -1 +0,0 @@
../offline-decode-files/WaveReader.cs

2
dotnet-examples/speaker-identification/speaker-identification.csproj
View File

@@ -9,7 +9,7 @@
</PropertyGroup>
<ItemGroup>
<PackageReference Include="org.k2fsa.sherpa.onnx" Version="*" />
<ProjectReference Include="..\Common\Common.csproj" />
</ItemGroup>
</Project>

6
dotnet-examples/speech-recognition-from-microphone/speech-recognition-from-microphone.csproj
View File

@@ -9,9 +9,11 @@
</PropertyGroup>
<ItemGroup>
<PackageReference Include="CommandLineParser" Version="2.9.1" />
<PackageReference Include="org.k2fsa.sherpa.onnx" Version="*" />
<PackageReference Include="PortAudioSharp2" Version="*" />
</ItemGroup>
<ItemGroup>
<ProjectReference Include="..\Common\Common.csproj" />
</ItemGroup>
</Project>

1
dotnet-examples/spoken-language-identification/WaveReader.cs
View File

@@ -1 +0,0 @@
../offline-decode-files/WaveReader.cs

2
dotnet-examples/spoken-language-identification/spoken-language-identification.csproj
View File

@@ -9,7 +9,7 @@
</PropertyGroup>
<ItemGroup>
<PackageReference Include="org.k2fsa.sherpa.onnx" Version="*" />
<ProjectReference Include="..\Common\Common.csproj" />
</ItemGroup>
</Project>

1
dotnet-examples/streaming-hlg-decoding/WaveReader.cs
View File

@@ -1 +0,0 @@
../online-decode-files/WaveReader.cs

2
dotnet-examples/streaming-hlg-decoding/streaming-hlg-decoding.csproj
View File

@@ -9,7 +9,7 @@
</PropertyGroup>
<ItemGroup>
<PackageReference Include="org.k2fsa.sherpa.onnx" Version="*" />
<ProjectReference Include="..\Common\Common.csproj" />
</ItemGroup>
</Project>

124
dotnet-examples/vad-non-streaming-asr-paraformer/Program.cs
View File

@@ -1,62 +1,62 @@
// 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();
}
}
}
}
}
// 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();
}
}
}
}
}

1
dotnet-examples/vad-non-streaming-asr-paraformer/WaveReader.cs
View File

@@ -1 +0,0 @@
../online-decode-files/WaveReader.cs

0
dotnet-examples/vad-non-streaming-asr-paraformer/run.sh Executable file → Normal file
View File

30
dotnet-examples/vad-non-streaming-asr-paraformer/vad-non-streaming-asr-paraformer.csproj
View File

@@ -1,15 +1,15 @@
<Project Sdk="Microsoft.NET.Sdk">
<PropertyGroup>
<OutputType>Exe</OutputType>
<TargetFramework>net6.0</TargetFramework>
<RootNamespace>vad_non_streaming_asr_paraformer</RootNamespace>
<ImplicitUsings>enable</ImplicitUsings>
<Nullable>enable</Nullable>
</PropertyGroup>
<ItemGroup>
<PackageReference Include="org.k2fsa.sherpa.onnx" Version="*" />
</ItemGroup>
</Project>
<Project Sdk="Microsoft.NET.Sdk">
<PropertyGroup>
<OutputType>Exe</OutputType>
<TargetFramework>net6.0</TargetFramework>
<RootNamespace>vad_non_streaming_asr_paraformer</RootNamespace>
<ImplicitUsings>enable</ImplicitUsings>
<Nullable>enable</Nullable>
</PropertyGroup>
<ItemGroup>
<ProjectReference Include="..\Common\Common.csproj" />
</ItemGroup>
</Project>