Add Go API for MatchaTTS models (#1685)

.github/workflows/test-go-package.yaml

@@ -209,6 +209,15 @@ jobs:
           go build
           ls -lh
 
+          echo "Test matcha zh"
+          ./run-matcha-zh.sh
+          rm -rf matcha-icefall-*
+
+          echo "Test matcha en"
+          ./run-matcha-en.sh
+          rm -rf matcha-icefall-*
+          ls -lh *.wav
+
           echo "Test vits-ljs"
           ./run-vits-ljs.sh
           rm -rf vits-ljs
@@ -246,6 +255,15 @@ jobs:
           cp -v /C/Users/runneradmin/go/pkg/mod/github.com/k2-fsa/sherpa-onnx-go-windows*/lib/x86_64-pc-windows-gnu/*.dll .
           ls -lh
 
+          echo "Test matcha zh"
+          ./run-matcha-zh.sh
+          rm -rf matcha-icefall-*
+
+          echo "Test matcha en"
+          ./run-matcha-en.sh
+          rm -rf matcha-icefall-*
+          ls -lh *.wav
+
           echo "Test vits-ljs"
           ./run-vits-ljs.sh
           rm -rf vits-ljs
@@ -291,6 +309,15 @@ jobs:
           cp -v /C/Users/runneradmin/go/pkg/mod/github.com/k2-fsa/sherpa-onnx-go-windows*/lib/i686-pc-windows-gnu/*.dll .
           ls -lh
 
+          echo "Test matcha zh"
+          ./run-matcha-zh.sh
+          rm -rf matcha-icefall-*
+
+          echo "Test matcha en"
+          ./run-matcha-en.sh
+          rm -rf matcha-icefall-*
+          ls -lh *.wav
+
           echo "Test vits-ljs"
           ./run-vits-ljs.sh
           rm -rf vits-ljs

.github/workflows/test-go.yaml

@@ -226,6 +226,15 @@ jobs:
           go build
           ls -lh
 
+          echo "Test matcha zh"
+          ./run-matcha-zh.sh
+          rm -rf matcha-icefall-*
+
+          echo "Test matcha en"
+          ./run-matcha-en.sh
+          rm -rf matcha-icefall-*
+          ls -lh *.wav
+
           echo "Test vits-ljs"
           ./run-vits-ljs.sh
           rm -rf vits-ljs

go-api-examples/non-streaming-tts/main.go

@@ -17,11 +17,22 @@ func main() {
 	flag.StringVar(&config.Model.Vits.Lexicon, "vits-lexicon", "", "Path to lexicon.txt")
 	flag.StringVar(&config.Model.Vits.Tokens, "vits-tokens", "", "Path to tokens.txt")
 	flag.StringVar(&config.Model.Vits.DataDir, "vits-data-dir", "", "Path to espeak-ng-data")
+	flag.StringVar(&config.Model.Matcha.DictDir, "vits-dict-dir", "", "Path to dict for jieba")
 
 	flag.Float32Var(&config.Model.Vits.NoiseScale, "vits-noise-scale", 0.667, "noise_scale for VITS")
 	flag.Float32Var(&config.Model.Vits.NoiseScaleW, "vits-noise-scale-w", 0.8, "noise_scale_w for VITS")
 	flag.Float32Var(&config.Model.Vits.LengthScale, "vits-length-scale", 1.0, "length_scale for VITS. small -> faster in speech speed; large -> slower")
 
+	flag.StringVar(&config.Model.Matcha.AcousticModel, "matcha-acoustic-model", "", "Path to the matcha acoustic model")
+	flag.StringVar(&config.Model.Matcha.Vocoder, "matcha-vocoder", "", "Path to the matcha vocoder model")
+	flag.StringVar(&config.Model.Matcha.Lexicon, "matcha-lexicon", "", "Path to lexicon.txt")
+	flag.StringVar(&config.Model.Matcha.Tokens, "matcha-tokens", "", "Path to tokens.txt")
+	flag.StringVar(&config.Model.Matcha.DataDir, "matcha-data-dir", "", "Path to espeak-ng-data")
+	flag.StringVar(&config.Model.Matcha.DictDir, "matcha-dict-dir", "", "Path to dict for jieba")
+
+	flag.Float32Var(&config.Model.Matcha.NoiseScale, "matcha-noise-scale", 0.667, "noise_scale for Matcha")
+	flag.Float32Var(&config.Model.Matcha.LengthScale, "matcha-length-scale", 1.0, "length_scale for Matcha. small -> faster in speech speed; large -> slower")
+
 	flag.IntVar(&config.Model.NumThreads, "num-threads", 1, "Number of threads for computing")
 	flag.IntVar(&config.Model.Debug, "debug", 0, "Whether to show debug message")
 	flag.StringVar(&config.Model.Provider, "provider", "cpu", "Provider to use")

go-api-examples/non-streaming-tts/run-matcha-en.sh

@@ -0,0 +1,31 @@
+#!/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
+
+go mod tidy
+go build
+
+./non-streaming-tts \
+  --matcha-acoustic-model=./matcha-icefall-en_US-ljspeech/model-steps-3.onnx \
+  --matcha-vocoder=./hifigan_v2.onnx \
+  --matcha-tokens=./matcha-icefall-en_US-ljspeech/tokens.txt \
+  --matcha-data-dir=./matcha-icefall-en_US-ljspeech/espeak-ng-data \
+  --debug=1 \
+  --output-filename=./test-matcha-en.wav \
+  "Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone."
+
+

go-api-examples/non-streaming-tts/run-matcha-zh.sh

@@ -0,0 +1,31 @@
+#!/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
+
+go mod tidy
+go build
+
+./non-streaming-tts \
+  --matcha-acoustic-model=./matcha-icefall-zh-baker/model-steps-3.onnx \
+  --matcha-vocoder=./hifigan_v2.onnx \
+  --matcha-lexicon=./matcha-icefall-zh-baker/lexicon.txt \
+  --matcha-tokens=./matcha-icefall-zh-baker/tokens.txt \
+  --matcha-dict-dir=./matcha-icefall-zh-baker/dict \
+  --debug=1 \
+  --tts-rule-fsts=./matcha-icefall-zh-baker/phone.fst,./matcha-icefall-zh-baker/date.fst,./matcha-icefall-zh-baker/number.fst \
+  --output-filename=./test-matcha-zh.wav \
+  "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。"
+

go-api-examples/non-streaming-tts/run-vits-piper-en_US-lessac-medium.sh

@@ -4,7 +4,7 @@ set -ex
 
 if [ ! -d vits-piper-en_US-lessac-medium ]; then
   curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_US-lessac-medium.tar.bz2
-  tar xvf vits-piper-en_US-lessac-medium.tar.bz2
+  tar xf vits-piper-en_US-lessac-medium.tar.bz2
   rm vits-piper-en_US-lessac-medium.tar.bz2
 fi
 

scripts/go/_internal/non-streaming-tts/run-matcha-en.sh

@@ -0,0 +1 @@
+../../../../go-api-examples/non-streaming-tts/run-matcha-en.sh

scripts/go/_internal/non-streaming-tts/run-matcha-zh.sh

@@ -0,0 +1 @@
+../../../../go-api-examples/non-streaming-tts/run-matcha-zh.sh

scripts/go/sherpa_onnx.go

@@ -671,8 +671,20 @@ type OfflineTtsVitsModelConfig struct {
 	DictDir     string  // Path to dict directory for jieba (used only in Chinese tts)
 }
 
+type OfflineTtsMatchaModelConfig struct {
+	AcousticModel string  // Path to the acoustic model for MatchaTTS
+	Vocoder       string  // Path to the vocoder model for MatchaTTS
+	Lexicon       string  // Path to lexicon.txt
+	Tokens        string  // Path to tokens.txt
+	DataDir       string  // Path to espeak-ng-data directory
+	NoiseScale    float32 // noise scale for vits models. Please use 0.667 in general
+	LengthScale   float32 // Please use 1.0 in general. Smaller -> Faster speech speed. Larger -> Slower speech speed
+	DictDir       string  // Path to dict directory for jieba (used only in Chinese tts)
+}
+
 type OfflineTtsModelConfig struct {
-	Vits OfflineTtsVitsModelConfig
+	Vits   OfflineTtsVitsModelConfig
+	Matcha OfflineTtsMatchaModelConfig
 
 	// Number of threads to use for neural network computation
 	NumThreads int
@@ -722,6 +734,7 @@ func NewOfflineTts(config *OfflineTtsConfig) *OfflineTts {
 
 	c.max_num_sentences = C.int(config.MaxNumSentences)
 
+	// vits
 	c.model.vits.model = C.CString(config.Model.Vits.Model)
 	defer C.free(unsafe.Pointer(c.model.vits.model))
 
@@ -741,6 +754,28 @@ func NewOfflineTts(config *OfflineTtsConfig) *OfflineTts {
 	c.model.vits.dict_dir = C.CString(config.Model.Vits.DictDir)
 	defer C.free(unsafe.Pointer(c.model.vits.dict_dir))
 
+	// matcha
+	c.model.matcha.acoustic_model = C.CString(config.Model.Matcha.AcousticModel)
+	defer C.free(unsafe.Pointer(c.model.matcha.acoustic_model))
+
+	c.model.matcha.vocoder = C.CString(config.Model.Matcha.Vocoder)
+	defer C.free(unsafe.Pointer(c.model.matcha.vocoder))
+
+	c.model.matcha.lexicon = C.CString(config.Model.Matcha.Lexicon)
+	defer C.free(unsafe.Pointer(c.model.matcha.lexicon))
+
+	c.model.matcha.tokens = C.CString(config.Model.Matcha.Tokens)
+	defer C.free(unsafe.Pointer(c.model.matcha.tokens))
+
+	c.model.matcha.data_dir = C.CString(config.Model.Matcha.DataDir)
+	defer C.free(unsafe.Pointer(c.model.matcha.data_dir))
+
+	c.model.matcha.noise_scale = C.float(config.Model.Matcha.NoiseScale)
+	c.model.matcha.length_scale = C.float(config.Model.Matcha.LengthScale)
+
+	c.model.matcha.dict_dir = C.CString(config.Model.Matcha.DictDir)
+	defer C.free(unsafe.Pointer(c.model.matcha.dict_dir))
+
 	c.model.num_threads = C.int(config.Model.NumThreads)
 	c.model.debug = C.int(config.Model.Debug)