This repository has been archived on 2025-08-26.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-mr_series-sherpa-onnx/dotnet-examples/offline-tts/Program.cs

206 lines
8.4 KiB
C#
Raw Normal View History

2024-01-28 23:29:39 +08:00
// Copyright (c) 2024 Xiaomi Corporation
//
// This file shows how to use a non-streaming TTS model for text-to-speech
// Please refer to
// https://k2-fsa.github.io/sherpa/onnx/pretrained_models/index.html
// and
// https://github.com/k2-fsa/sherpa-onnx/releases/tag/tts-models
// to download pre-trained models
using CommandLine;
using CommandLine.Text;
2024-01-28 23:29:39 +08:00
using SherpaOnnx;
class OfflineTtsDemo
{
class Options
{
[Option("tts-rule-fsts", Required = false, Default = "", HelpText = "path to rule.fst")]
public string RuleFsts { get; set; } = string.Empty;
2024-01-28 23:29:39 +08:00
[Option("tts-rule-fars", Required = false, Default = "", HelpText = "path to rule.far")]
public string RuleFars { get; set; } = string.Empty;
[Option("dict-dir", Required = false, Default = "", HelpText = "Path to the directory containing dict for jieba.")]
public string DictDir { get; set; } = string.Empty;
[Option("data-dir", Required = false, Default = "", HelpText = "Path to the directory containing dict for espeak-ng.")]
public string DataDir { get; set; } = string.Empty;
2024-01-28 23:29:39 +08:00
[Option("length-scale", Required = false, Default = 1, HelpText = "speech speed. Larger->Slower; Smaller->faster")]
2024-06-28 11:27:19 +08:00
public float LengthScale { get; set; } = 1;
2024-01-28 23:29:39 +08:00
[Option("noise-scale", Required = false, Default = 0.667f, HelpText = "noise_scale for VITS or Matcha models")]
2024-06-28 11:27:19 +08:00
public float NoiseScale { get; set; } = 0.667F;
2024-01-28 23:29:39 +08:00
2024-06-28 11:27:19 +08:00
[Option("vits-noise-scale-w", Required = false, Default = 0.8F, HelpText = "noise_scale_w for VITS models")]
public float NoiseScaleW { get; set; } = 0.8F;
2024-01-28 23:29:39 +08:00
[Option("lexicon", Required = false, Default = "", HelpText = "Path to lexicon.txt")]
public string Lexicon { get; set; } = string.Empty;
2024-01-28 23:29:39 +08:00
[Option("tokens", Required = true, Default = "", HelpText = "Path to tokens.txt")]
public string Tokens { get; set; } = string.Empty;
2024-01-28 23:29:39 +08:00
[Option("tts-max-num-sentences", Required = false, Default = 1, HelpText = "Maximum number of sentences that we process at a time.")]
2024-06-28 11:27:19 +08:00
public int MaxNumSentences { get; set; } = 1;
2024-01-28 23:29:39 +08:00
[Option(Required = false, Default = 0, HelpText = "1 to show debug messages.")]
2024-06-28 11:27:19 +08:00
public int Debug { get; set; } = 0;
2024-01-28 23:29:39 +08:00
[Option("vits-model", Required = false, HelpText = "Path to VITS model")]
public string Model { get; set; } = string.Empty;
2024-01-28 23:29:39 +08:00
[Option("matcha-acoustic-model", Required = false, HelpText = "Path to the acoustic model of Matcha")]
public string AcousticModel { get; set; } = "";
[Option("matcha-vocoder", Required = false, HelpText = "Path to the vocoder model of Matcha")]
public string Vocoder { get; set; } = "";
2024-01-28 23:29:39 +08:00
[Option("sid", Required = false, Default = 0, HelpText = "Speaker ID")]
2024-06-28 11:27:19 +08:00
public int SpeakerId { get; set; } = 0;
2024-01-28 23:29:39 +08:00
[Option("text", Required = true, HelpText = "Text to synthesize")]
public string Text { get; set; } = string.Empty;
2024-01-28 23:29:39 +08:00
[Option("output-filename", Required = true, Default = "./generated.wav", HelpText = "Path to save the generated audio")]
2024-06-28 11:27:19 +08:00
public string OutputFilename { get; set; } = "./generated.wav";
2024-01-28 23:29:39 +08:00
}
static void Main(string[] args)
{
var parser = new Parser(with => with.HelpWriter = null);
2024-01-28 23:29:39 +08:00
var parserResult = parser.ParseArguments<Options>(args);
parserResult
.WithParsed<Options>(options => Run(options))
.WithNotParsed(errs => DisplayHelp(parserResult, errs));
}
private static void DisplayHelp<T>(ParserResult<T> result, IEnumerable<Error> errs)
{
var usage = @"
# matcha-icefall-zh-baker
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/matcha-icefall-zh-baker.tar.bz2
tar xvf matcha-icefall-zh-baker.tar.bz2
rm matcha-icefall-zh-baker.tar.bz2
2025-03-17 17:05:15 +08:00
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
dotnet run \
--matcha-acoustic-model=./matcha-icefall-zh-baker/model-steps-3.onnx \
2025-03-17 17:05:15 +08:00
--matcha-vocoder=./vocos-22khz-univ.onnx \
--lexicon=./matcha-icefall-zh-baker/lexicon.txt \
--tokens=./matcha-icefall-zh-baker/tokens.txt \
--dict-dir=./matcha-icefall-zh-baker/dict \
--tts-rule-fsts=./matcha-icefall-zh-baker/phone.fst,./matcha-icefall-zh-baker/date.fst,./matcha-icefall-zh-baker/number.fst \
--debug=1 \
--output-filename=./matcha-zh.wav \
--text='; 2024123111018920240511123456'
# matcha-icefall-en_US-ljspeech
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/matcha-icefall-en_US-ljspeech.tar.bz2
tar xvf matcha-icefall-en_US-ljspeech.tar.bz2
rm matcha-icefall-en_US-ljspeech.tar.bz2
2025-03-17 17:05:15 +08:00
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
dotnet run \
--matcha-acoustic-model=./matcha-icefall-en_US-ljspeech/model-steps-3.onnx \
2025-03-17 17:05:15 +08:00
--matcha-vocoder=./vocos-22khz-univ.onnx \
--tokens=./matcha-icefall-zh-baker/tokens.txt \
--data-dir=./matcha-icefall-en_US-ljspeech/espeak-ng-data \
--debug=1 \
--output-filename=./matcha-zh.wav \
--text='Today as always, men fall into two groups: slaves and free men. Whoever does not have two-thirds of his day for himself, is a slave, whatever he may be: a statesman, a businessman, an official, or a scholar.'
2024-01-28 23:29:39 +08:00
# vits-aishell3
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-icefall-zh-aishell3.tar.bz2
tar xvf vits-icefall-zh-aishell3.tar.bz2
2024-01-28 23:29:39 +08:00
dotnet run \
--vits-model=./vits-icefall-zh-aishell3/model.onnx \
--tokens=./vits-icefall-zh-aishell3/tokens.txt \
--lexicon=./vits-icefall-zh-aishell3/lexicon.txt \
--tts-rule-fsts=./vits-icefall-zh-aishell3/phone.fst,./vits-icefall-zh-aishell3/date.fst,./vits-icefall-zh-aishell3/number.fst \
--tts-rule-fars=./vits-icefall-zh-aishell3/rule.far \
2024-01-28 23:29:39 +08:00
--sid=66 \
--debug=1 \
--output-filename=./aishell3-66.wav \
--text=
# Piper models
wget -qq https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-amy-low.tar.bz2
tar xf vits-piper-en_US-amy-low.tar.bz2
dotnet run \
--vits-model=./vits-piper-en_US-amy-low/en_US-amy-low.onnx \
--tokens=./vits-piper-en_US-amy-low/tokens.txt \
--data-dir=./vits-piper-en_US-amy-low/espeak-ng-data \
2024-01-28 23:29:39 +08:00
--debug=1 \
--output-filename=./amy.wav \
--text='This is a text to speech application in dotnet with Next Generation Kaldi'
Please refer to
https://k2-fsa.github.io/sherpa/onnx/tts/pretrained_models/index.html
to download more models.
";
var helpText = HelpText.AutoBuild(result, h =>
{
h.AdditionalNewLineAfterOption = false;
h.Heading = usage;
h.Copyright = "Copyright (c) 2024 Xiaomi Corporation";
return HelpText.DefaultParsingErrorsHandler(result, h);
}, e => e);
Console.WriteLine(helpText);
}
private static void Run(Options options)
{
var config = new OfflineTtsConfig();
2024-01-28 23:29:39 +08:00
config.Model.Vits.Model = options.Model;
config.Model.Vits.Lexicon = options.Lexicon;
config.Model.Vits.Tokens = options.Tokens;
config.Model.Vits.DataDir = options.DataDir;
config.Model.Vits.DictDir = options.DictDir;
2024-01-28 23:29:39 +08:00
config.Model.Vits.NoiseScale = options.NoiseScale;
config.Model.Vits.NoiseScaleW = options.NoiseScaleW;
config.Model.Vits.LengthScale = options.LengthScale;
config.Model.Matcha.AcousticModel = options.AcousticModel;
config.Model.Matcha.Vocoder = options.Vocoder;
config.Model.Matcha.Lexicon = options.Lexicon;
config.Model.Matcha.Tokens = options.Tokens;
config.Model.Matcha.DataDir = options.DataDir;
config.Model.Matcha.DictDir = options.DictDir;
config.Model.Matcha.NoiseScale = options.NoiseScale;
config.Model.Matcha.LengthScale = options.LengthScale;
2024-01-28 23:29:39 +08:00
config.Model.NumThreads = 1;
config.Model.Debug = options.Debug;
config.Model.Provider = "cpu";
config.RuleFsts = options.RuleFsts;
config.RuleFars = options.RuleFars;
2024-01-28 23:29:39 +08:00
config.MaxNumSentences = options.MaxNumSentences;
var tts = new OfflineTts(config);
var speed = 1.0f / options.LengthScale;
var sid = options.SpeakerId;
var audio = tts.Generate(options.Text, speed, sid);
var ok = audio.SaveToWaveFile(options.OutputFilename);
2024-01-28 23:29:39 +08:00
if (ok)
{
Console.WriteLine($"Wrote to {options.OutputFilename} succeeded!");
}
else
{
Console.WriteLine($"Failed to write {options.OutputFilename}");
}
}
}