Add C# and JavaScript (wasm) API for MatchaTTS models

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

@@ -2,7 +2,27 @@
 
 cd dotnet-examples/
 
-cd ./offline-speaker-diarization
+cd ./offline-tts
+./run-matcha-zh.sh
+ls -lh *.wav
+./run-matcha-en.sh
+ls -lh *.wav
+./run-aishell3.sh
+ls -lh *.wav
+./run-piper.sh
+ls -lh *.wav
+./run-hf-fanchen.sh
+ls -lh *.wav
+ls -lh
+
+pushd ../..
+
+mkdir tts
+
+cp dotnet-examples/offline-tts/*.wav ./tts
+popd
+
+cd ../offline-speaker-diarization
 ./run.sh
 rm -rfv *.onnx
 rm -fv *.wav
@@ -76,14 +96,4 @@ cd ../spoken-language-identification
 ./run.sh
 rm -rf sherpa-onnx-*
 
-cd ../offline-tts
-./run-aishell3.sh
-./run-piper.sh
-./run-hf-fanchen.sh
-ls -lh
 
-cd ../..
-
-mkdir tts
-
-cp dotnet-examples/offline-tts/*.wav ./tts

.github/scripts/test-nodejs-npm.sh

@@ -9,6 +9,48 @@ git status
 ls -lh
 ls -lh node_modules
 
+# offline tts
+#
+curl -SL -O 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
+
+curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
+
+node ./test-offline-tts-matcha-zh.js
+
+rm -rf matcha-icefall-zh-baker
+rm hifigan_v2.onnx
+
+echo "---"
+
+curl -SL -O 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
+
+wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
+
+node ./test-offline-tts-matcha-en.js
+
+rm -rf matcha-icefall-en_US-ljspeech
+rm hifigan_v2.onnx
+
+echo "---"
+
+curl -LS -O 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
+node ./test-offline-tts-vits-en.js
+rm -rf vits-piper-en_US-amy-low*
+
+echo "---"
+
+curl -LS -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
+node ./test-offline-tts-vits-zh.js
+rm -rf vits-icefall-zh-aishell3*
+
+ls -lh *.wav
+
 echo '-----speaker diarization----------'
 curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/speaker-segmentation-models/sherpa-onnx-pyannote-segmentation-3-0.tar.bz2
 tar xvf sherpa-onnx-pyannote-segmentation-3-0.tar.bz2
@@ -147,15 +189,3 @@ tar xvf sherpa-onnx-streaming-zipformer-ctc-small-2024-03-18.tar.bz2
 rm sherpa-onnx-streaming-zipformer-ctc-small-2024-03-18.tar.bz2
 node ./test-online-zipformer2-ctc-hlg.js
 rm -rf sherpa-onnx-streaming-zipformer-ctc-small-2024-03-18
-
-# offline tts
-
-curl -LS -O 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
-node ./test-offline-tts-en.js
-rm -rf vits-piper-en_US-amy-low*
-
-curl -LS -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
-node ./test-offline-tts-zh.js
-rm -rf vits-icefall-zh-aishell3*

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

@@ -92,6 +92,50 @@ jobs:
         python-version: ["3.8"]
 
     steps:
+      - name: Check space
+        shell: bash
+        run: |
+          df -h
+
+      - name: Free space
+        shell: bash
+        run: |
+          df -h
+          rm -rf /opt/hostedtoolcache
+          df -h
+
+      - name: Free more space
+        shell: bash
+        run: |
+          # https://github.com/orgs/community/discussions/25678
+          cd /opt
+          find . -maxdepth 1 -mindepth 1 '!' -path ./containerd '!' -path ./actionarchivecache '!' -path ./runner '!' -path ./runner-cache -exec rm -rf '{}' ';'
+
+          sudo rm -rf /usr/share/dotnet
+          sudo rm -rf "/usr/local/share/boost"
+          sudo rm -rf "$AGENT_TOOLSDIRECTORY"
+
+      - name: Free Disk Space (Ubuntu)
+        uses: jlumbroso/free-disk-space@main
+        with:
+          # this might remove tools that are actually needed,
+          # if set to "true" but frees about 6 GB
+          tool-cache: false
+
+          # all of these default to true, but feel free to set to
+          # "false" if necessary for your workflow
+          android: true
+          dotnet: false
+          haskell: true
+          large-packages: true
+          docker-images: false
+          swap-storage: true
+
+      - name: Check space
+        shell: bash
+        run: |
+          df -h
+
       - uses: actions/checkout@v4
         with:
           fetch-depth: 0

dotnet-examples/offline-tts-play/Program.cs

@@ -21,48 +21,56 @@ class OfflineTtsPlayDemo
 {
   class Options
   {
-
     [Option("tts-rule-fsts", Required = false, Default = "", HelpText = "path to rule.fst")]
-    public string? RuleFsts { get; set; }
+    public string RuleFsts { get; set; } = string.Empty;
+
+    [Option("tts-rule-fars", Required = false, Default = "", HelpText = "path to rule.far")]
+    public string RuleFars { get; set; } = string.Empty;
 
-    [Option("vits-dict-dir", Required = false, Default = "", HelpText = "Path to the directory containing dict for jieba.")]
-    public string? DictDir { get; set; }
+    [Option("dict-dir", Required = false, Default = "", HelpText = "Path to the directory containing dict for jieba.")]
+    public string DictDir { get; set; } = string.Empty;
 
-    [Option("vits-data-dir", Required = false, Default = "", HelpText = "Path to the directory containing dict for espeak-ng.")]
-    public string? DataDir { get; set; }
+    [Option("data-dir", Required = false, Default = "", HelpText = "Path to the directory containing dict for espeak-ng.")]
+    public string DataDir { get; set; } = string.Empty;
 
-    [Option("vits-length-scale", Required = false, Default = 1, HelpText = "speech speed. Larger->Slower; Smaller->faster")]
-    public float LengthScale { get; set; }
+    [Option("length-scale", Required = false, Default = 1, HelpText = "speech speed. Larger->Slower; Smaller->faster")]
+    public float LengthScale { get; set; } = 1;
 
-    [Option("vits-noise-scale", Required = false, Default = 0.667f, HelpText = "noise_scale for VITS models")]
-    public float NoiseScale { get; set; }
+    [Option("noise-scale", Required = false, Default = 0.667f, HelpText = "noise_scale for VITS or Matcha models")]
+    public float NoiseScale { get; set; } = 0.667F;
 
-    [Option("vits-noise-scale-w", Required = false, Default = 0.8f, HelpText = "noise_scale_w for VITS models")]
-    public float NoiseScaleW { get; set; }
+    [Option("vits-noise-scale-w", Required = false, Default = 0.8F, HelpText = "noise_scale_w for VITS models")]
+    public float NoiseScaleW { get; set; } = 0.8F;
 
-    [Option("vits-lexicon", Required = false, Default = "", HelpText = "Path to lexicon.txt")]
-    public string? Lexicon { get; set; }
+    [Option("lexicon", Required = false, Default = "", HelpText = "Path to lexicon.txt")]
+    public string Lexicon { get; set; } = string.Empty;
 
-    [Option("vits-tokens", Required = false, Default = "", HelpText = "Path to tokens.txt")]
-    public string? Tokens { get; set; }
+    [Option("tokens", Required = true, Default = "", HelpText = "Path to tokens.txt")]
+    public string Tokens { get; set; } = string.Empty;
 
     [Option("tts-max-num-sentences", Required = false, Default = 1, HelpText = "Maximum number of sentences that we process at a time.")]
-    public int MaxNumSentences { get; set; }
+    public int MaxNumSentences { get; set; } = 1;
 
     [Option(Required = false, Default = 0, HelpText = "1 to show debug messages.")]
-    public int Debug { get; set; }
+    public int Debug { get; set; } = 0;
+
+    [Option("vits-model", Required = false, HelpText = "Path to VITS model")]
+    public string Model { get; set; } = string.Empty;
 
-    [Option("vits-model", Required = true, HelpText = "Path to VITS model")]
-    public string? Model { get; set; }
+    [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; } = "";
 
     [Option("sid", Required = false, Default = 0, HelpText = "Speaker ID")]
-    public int SpeakerId { get; set; }
+    public int SpeakerId { get; set; } = 0;
 
     [Option("text", Required = true, HelpText = "Text to synthesize")]
-    public string? Text { get; set; }
+    public string Text { get; set; } = string.Empty;
 
     [Option("output-filename", Required = true, Default = "./generated.wav", HelpText = "Path to save the generated audio")]
-    public string? OutputFilename { get; set; }
+    public string OutputFilename { get; set; } = "./generated.wav";
   }
 
   static void Main(string[] args)
@@ -78,15 +86,51 @@ static void Main(string[] args)
   private static void DisplayHelp<T>(ParserResult<T> result, IEnumerable<Error> errs)
   {
     string 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
+
+wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
+
+dotnet run \
+  --matcha-acoustic-model=./matcha-icefall-zh-baker/model-steps-3.onnx \
+  --matcha-vocoder=./hifigan_v2.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='某某银行的副行长和一些行政领导表示，他们去过长江和长白山; 经济不断增长。2024年12月31号，拨打110或者18920240511。123456块钱。'
+
+# 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
+
+wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
+
+dotnet run \
+  --matcha-acoustic-model=./matcha-icefall-en_US-ljspeech/model-steps-3.onnx \
+  --matcha-vocoder=./hifigan_v2.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.'
+
 # vits-aishell3
 
 wget -qq https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-zh-aishell3.tar.bz2
 tar xf vits-zh-aishell3.tar.bz2
 
 dotnet run \
   --vits-model=./vits-zh-aishell3/vits-aishell3.onnx \
-  --vits-tokens=./vits-zh-aishell3/tokens.txt \
-  --vits-lexicon=./vits-zh-aishell3/lexicon.txt \
+  --tokens=./vits-zh-aishell3/tokens.txt \
+  --lexicon=./vits-zh-aishell3/lexicon.txt \
   --tts-rule-fsts=./vits-zh-aishell3/rule.fst \
   --sid=66 \
   --debug=1 \
@@ -100,8 +144,8 @@ 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 \
-  --vits-tokens=./vits-piper-en_US-amy-low/tokens.txt \
-  --vits-data-dir=./vits-piper-en_US-amy-low/espeak-ng-data \
+  ---tokens=./vits-piper-en_US-amy-low/tokens.txt \
+  --data-dir=./vits-piper-en_US-amy-low/espeak-ng-data \
   --debug=1 \
   --output-filename=./amy.wav \
   --text='This is a text to speech application in dotnet with Next Generation Kaldi'
@@ -124,6 +168,7 @@ to download more models.
   private static void Run(Options options)
   {
     var config = new OfflineTtsConfig();
+
     config.Model.Vits.Model = options.Model;
     config.Model.Vits.Lexicon = options.Lexicon;
     config.Model.Vits.Tokens = options.Tokens;
@@ -132,6 +177,16 @@ private static void Run(Options options)
     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;
+
     config.Model.NumThreads = 1;
     config.Model.Debug = options.Debug;
     config.Model.Provider = "cpu";

dotnet-examples/offline-tts-play/run-hf-fanchen.sh

@@ -8,8 +8,8 @@ fi
 
 dotnet run \
   --vits-model=./vits-zh-hf-fanchen-C/vits-zh-hf-fanchen-C.onnx \
-  --vits-tokens=./vits-zh-hf-fanchen-C/tokens.txt \
-  --vits-lexicon=./vits-zh-hf-fanchen-C/lexicon.txt \
+  --tokens=./vits-zh-hf-fanchen-C/tokens.txt \
+  --lexicon=./vits-zh-hf-fanchen-C/lexicon.txt \
   --tts-rule-fsts=./vits-zh-hf-fanchen-C/phone.fst,./vits-zh-hf-fanchen-C/date.fst,./vits-zh-hf-fanchen-C/number.fst \
   --vits-dict-dir=./vits-zh-hf-fanchen-C/dict \
   --sid=100 \

dotnet-examples/offline-tts-play/run-matcha-en.sh

@@ -0,0 +1,26 @@
+#!/usr/bin/env bash
+set -ex
+
+
+# please visit
+# https://k2-fsa.github.io/sherpa/onnx/tts/pretrained_models/matcha.html#matcha-icefall-en-us-ljspeech-american-english-1-female-speaker
+# matcha.html#matcha-icefall-en-us-ljspeech-american-english-1-female-speaker
+# to download more models
+if [ ! -f ./matcha-icefall-en_US-ljspeech/model-steps-3.onnx ]; then
+  curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/matcha-icefall-en_US-ljspeech.tar.bz2
+  tar xf matcha-icefall-en_US-ljspeech.tar.bz2
+  rm matcha-icefall-en_US-ljspeech.tar.bz2
+fi
+
+if [ ! -f ./hifigan_v2.onnx ]; then
+  curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
+fi
+
+dotnet run \
+  --matcha-acoustic-model=./matcha-icefall-en_US-ljspeech/model-steps-3.onnx \
+  --matcha-vocoder=./hifigan_v2.onnx \
+  --tokens=./matcha-icefall-en_US-ljspeech/tokens.txt \
+  --data-dir=./matcha-icefall-en_US-ljspeech/espeak-ng-data \
+  --debug=1 \
+  --output-filename=./matcha-en.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.'

dotnet-examples/offline-tts-play/run-matcha-zh.sh

@@ -0,0 +1,27 @@
+#!/usr/bin/env bash
+set -ex
+
+# please visit
+# https://k2-fsa.github.io/sherpa/onnx/tts/pretrained_models/matcha.html#matcha-icefall-zh-baker-chinese-1-female-speaker
+# to download more models
+if [ ! -f ./matcha-icefall-zh-baker/model-steps-3.onnx ]; then
+  curl -SL -O 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
+fi
+
+if [ ! -f ./hifigan_v2.onnx ]; then
+  curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
+fi
+
+
+dotnet run \
+  --matcha-acoustic-model=./matcha-icefall-zh-baker/model-steps-3.onnx \
+  --matcha-vocoder=./hifigan_v2.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="某某银行的副行长和一些行政领导表示，他们去过长江和长白山; 经济不断增长。2024年12月31号，拨打110或者18920240511。123456块钱。"