Add dict_dir arg to c api to support Chinese TTS models using jieba (#809)

CMakeLists.txt

@@ -1,7 +1,7 @@
 cmake_minimum_required(VERSION 3.13 FATAL_ERROR)
 project(sherpa-onnx)
 
-set(SHERPA_ONNX_VERSION "1.9.22")
+set(SHERPA_ONNX_VERSION "1.9.23")
 
 # Disable warning about
 #

mfc-examples/NonStreamingTextToSpeech/NonStreamingTextToSpeechDlg.cpp

@@ -470,6 +470,19 @@ void CNonStreamingTextToSpeechDlg::Init() {
   } else if (Exists("./lexicon.txt")) {
     config.model.vits.lexicon = "./lexicon.txt";
   }
+
+  if (Exists("./dict/jieba.dict.utf8")) {
+    config.model.vits.dict_dir = "./dict";
+  }
+
+  if (Exists("./phone.fst") && Exists("./date.fst") && Exists("./number.fst")) {
+    config.rule_fsts = "./phone.fst,./date.fst,number.fst";
+  }
+
+  if (Exists("./rule.far")) {
+    config.rule_fars = "./rule.far";
+  }
+
   config.model.vits.tokens = "./tokens.txt";
 
   tts_ = SherpaOnnxCreateOfflineTts(&config);

nodejs-examples/test-offline-tts-en.js

@@ -8,6 +8,7 @@ function createOfflineTts() {
     lexicon: '',
     tokens: './vits-piper-en_US-amy-low/tokens.txt',
     dataDir: './vits-piper-en_US-amy-low/espeak-ng-data',
+    dictDir: '',
     noiseScale: 0.667,
     noiseScaleW: 0.8,
     lengthScale: 1.0,

nodejs-examples/test-offline-tts-zh.js

@@ -8,6 +8,7 @@ function createOfflineTts() {
     lexicon: './vits-icefall-zh-aishell3/lexicon.txt',
     tokens: './vits-icefall-zh-aishell3/tokens.txt',
     dataDir: '',
+    dictDir: '',
     noiseScale: 0.667,
     noiseScaleW: 0.8,
     lengthScale: 1.0,

scripts/dotnet/offline.cs

@@ -23,6 +23,8 @@ namespace SherpaOnnx
       NoiseScale = 0.667F;
       NoiseScaleW = 0.8F;
       LengthScale = 1.0F;
+
+      DictDir = "";
     }
     [MarshalAs(UnmanagedType.LPStr)]
     public string Model;
@@ -39,6 +41,9 @@ namespace SherpaOnnx
     public float NoiseScale;
     public float NoiseScaleW;
     public float LengthScale;
+
+    [MarshalAs(UnmanagedType.LPStr)]
+    public string DictDir;
   }
 
   [StructLayout(LayoutKind.Sequential)]

scripts/go/sherpa_onnx.go

@@ -532,10 +532,11 @@ type OfflineTtsVitsModelConfig struct {
 	Model       string  // Path to the VITS onnx model
 	Lexicon     string  // Path to lexicon.txt
 	Tokens      string  // Path to tokens.txt
-	DataDir     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
 	NoiseScaleW float32 // noise scale for vits models. Please use 0.8 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 {
@@ -605,6 +606,9 @@ func NewOfflineTts(config *OfflineTtsConfig) *OfflineTts {
 	c.model.vits.noise_scale_w = C.float(config.Model.Vits.NoiseScaleW)
 	c.model.vits.length_scale = C.float(config.Model.Vits.LengthScale)
 
+	c.model.vits.dict_dir = C.CString(config.Model.Vits.DictDir)
+	defer C.free(unsafe.Pointer(c.model.vits.dict_dir))
+
 	c.model.num_threads = C.int(config.Model.NumThreads)
 	c.model.debug = C.int(config.Model.Debug)
 

sherpa-onnx/c-api/c-api.cc

@@ -818,6 +818,8 @@ SherpaOnnxOfflineTts *SherpaOnnxCreateOfflineTts(
       SHERPA_ONNX_OR(config->model.vits.noise_scale_w, 0.8);
   tts_config.model.vits.length_scale =
       SHERPA_ONNX_OR(config->model.vits.length_scale, 1.0);
+  tts_config.model.vits.dict_dir =
+      SHERPA_ONNX_OR(config->model.vits.dict_dir, "");
 
   tts_config.model.num_threads = SHERPA_ONNX_OR(config->model.num_threads, 1);
   tts_config.model.debug = config->model.debug;

sherpa-onnx/c-api/c-api.h

@@ -772,6 +772,7 @@ SHERPA_ONNX_API typedef struct SherpaOnnxOfflineTtsVitsModelConfig {
   float noise_scale;
   float noise_scale_w;
   float length_scale;  // < 1, faster in speed; > 1, slower in speed
+  const char *dict_dir;
 } SherpaOnnxOfflineTtsVitsModelConfig;
 
 SHERPA_ONNX_API typedef struct SherpaOnnxOfflineTtsModelConfig {

swift-api-examples/SherpaOnnx.swift

@@ -623,7 +623,8 @@ func sherpaOnnxOfflineTtsVitsModelConfig(
   dataDir: String = "",
   noiseScale: Float = 0.667,
   noiseScaleW: Float = 0.8,
-  lengthScale: Float = 1.0
+  lengthScale: Float = 1.0,
+  dictDir: String = ""
 ) -> SherpaOnnxOfflineTtsVitsModelConfig {
   return SherpaOnnxOfflineTtsVitsModelConfig(
     model: toCPointer(model),
@@ -632,7 +633,8 @@ func sherpaOnnxOfflineTtsVitsModelConfig(
     data_dir: toCPointer(dataDir),
     noise_scale: noiseScale,
     noise_scale_w: noiseScaleW,
-    length_scale: lengthScale)
+    length_scale: lengthScale,
+    dict_dir: toCPointer(dictDir))
 }
 
 func sherpaOnnxOfflineTtsModelConfig(

wasm/nodejs/sherpa-onnx-wasm-nodejs.cc

@@ -43,6 +43,7 @@ void PrintOfflineTtsConfig(SherpaOnnxOfflineTtsConfig *tts_config) {
   fprintf(stdout, "noise scale: %.3f\n", vits_model_config->noise_scale);
   fprintf(stdout, "noise scale w: %.3f\n", vits_model_config->noise_scale_w);
   fprintf(stdout, "length scale: %.3f\n", vits_model_config->length_scale);
+  fprintf(stdout, "dict_dir: %s\n", vits_model_config->dict_dir);
 
   fprintf(stdout, "----------tts model config----------\n");
   fprintf(stdout, "num threads: %d\n", tts_model_config->num_threads);

wasm/tts/sherpa-onnx-tts.js

@@ -18,7 +18,12 @@ function initSherpaOnnxOfflineTtsVitsModelConfig(config, Module) {
   const tokensLen = Module.lengthBytesUTF8(config.tokens) + 1;
   const dataDirLen = Module.lengthBytesUTF8(config.dataDir) + 1;
 
-  const n = modelLen + lexiconLen + tokensLen + dataDirLen;
+  if (!('dictDir' in config)) {
+    config.dictDir = ''
+  }
+  const dictDirLen = Module.lengthBytesUTF8(config.dictDir) + 1;
+
+  const n = modelLen + lexiconLen + tokensLen + dataDirLen + dictDirLen;
 
   const buffer = Module._malloc(n);
 
@@ -38,6 +43,9 @@ function initSherpaOnnxOfflineTtsVitsModelConfig(config, Module) {
   Module.stringToUTF8(config.dataDir, buffer + offset, dataDirLen);
   offset += dataDirLen;
 
+  Module.stringToUTF8(config.dictDir, buffer + offset, dictDirLen);
+  offset += dictDirLen;
+
   offset = 0;
   Module.setValue(ptr, buffer + offset, 'i8*');
   offset += modelLen;
@@ -54,6 +62,8 @@ function initSherpaOnnxOfflineTtsVitsModelConfig(config, Module) {
   Module.setValue(ptr + 16, config.noiseScale, 'float');
   Module.setValue(ptr + 20, config.noiseScaleW, 'float');
   Module.setValue(ptr + 24, config.lengthScale, 'float');
+  Module.setValue(ptr + 28, buffer + offset, 'i8*');
+  offset += dictDirLen;
 
   return {
     buffer: buffer, ptr: ptr, len: len,
@@ -184,6 +194,7 @@ function createOfflineTts(Module, myConfig) {
     lexicon: '',
     tokens: './tokens.txt',
     dataDir: './espeak-ng-data',
+    dictDir: '',
     noiseScale: 0.667,
     noiseScaleW: 0.8,
     lengthScale: 1.0,

wasm/tts/sherpa-onnx-wasm-main-tts.cc

@@ -13,7 +13,7 @@
 
 extern "C" {
 
-static_assert(sizeof(SherpaOnnxOfflineTtsVitsModelConfig) == 7 * 4, "");
+static_assert(sizeof(SherpaOnnxOfflineTtsVitsModelConfig) == 8 * 4, "");
 static_assert(sizeof(SherpaOnnxOfflineTtsModelConfig) ==
                   sizeof(SherpaOnnxOfflineTtsVitsModelConfig) + 3 * 4,
               "");
@@ -32,6 +32,7 @@ void MyPrint(SherpaOnnxOfflineTtsConfig *tts_config) {
   fprintf(stdout, "noise scale: %.3f\n", vits_model_config->noise_scale);
   fprintf(stdout, "noise scale w: %.3f\n", vits_model_config->noise_scale_w);
   fprintf(stdout, "length scale: %.3f\n", vits_model_config->length_scale);
+  fprintf(stdout, "dict_dir: %s\n", vits_model_config->dict_dir);
 
   fprintf(stdout, "----------tts model config----------\n");
   fprintf(stdout, "num threads: %d\n", tts_model_config->num_threads);