add speed parameter

cpp/README.md

@@ -4,6 +4,8 @@ High-performance text-to-speech inference using ONNX Runtime.
 
 ## 📰 Update News
 
+**2025.11.19** - Added `--speed` parameter to control speech synthesis speed (default: 1.05, recommended range: 0.9-1.5).
+
 **2025.11.19** - Added automatic text chunking for long-form inference. Long texts are split into chunks and synthesized with natural pauses.
 
 ## Requirements

cpp/example_onnx.cpp

@@ -10,6 +10,7 @@ namespace fs = std::filesystem;
 struct Args {
     std::string onnx_dir = "../assets/onnx";
     int total_step = 5;
+    float speed = 1.05f;
     int n_test = 4;
     std::vector<std::string> voice_style = {"../assets/voice_styles/M1.json"};
     std::vector<std::string> text = {
@@ -36,6 +37,7 @@ Args parseArgs(int argc, char* argv[]) {
         std::string arg = argv[i];
         if (arg == "--onnx-dir" && i + 1 < argc) args.onnx_dir = argv[++i];
         else if (arg == "--total-step" && i + 1 < argc) args.total_step = std::stoi(argv[++i]);
+        else if (arg == "--speed" && i + 1 < argc) args.speed = std::stof(argv[++i]);
         else if (arg == "--n-test" && i + 1 < argc) args.n_test = std::stoi(argv[++i]);
         else if (arg == "--voice-style" && i + 1 < argc) args.voice_style = splitString(argv[++i], ',');
         else if (arg == "--text" && i + 1 < argc) args.text = splitString(argv[++i], '|');
@@ -51,6 +53,7 @@ int main(int argc, char* argv[]) {
     // --- 1. Parse arguments --- //
     Args args = parseArgs(argc, argv);
     int total_step = args.total_step;
+    float speed = args.speed;
     int n_test = args.n_test;
     std::string save_dir = args.save_dir;
     std::vector<std::string> voice_style_paths = args.voice_style;
@@ -84,9 +87,9 @@ int main(int argc, char* argv[]) {
         
         auto result = timer("Generating speech from text", [&]() {
             if (batch) {
-                return text_to_speech->batch(memory_info, text_list, style, total_step);
+                return text_to_speech->batch(memory_info, text_list, style, total_step, speed);
             } else {
-                return text_to_speech->call(memory_info, text_list[0], style, total_step);
+                return text_to_speech->call(memory_info, text_list[0], style, total_step, speed);
             }
         });
         

cpp/helper.cpp

@@ -160,7 +160,8 @@ TextToSpeech::SynthesisResult TextToSpeech::_infer(
     Ort::MemoryInfo& memory_info,
     const std::vector<std::string>& text_list,
     const Style& style,
-    int total_step
+    int total_step,
+    float speed
 ) {
     int bsz = text_list.size();
     
@@ -213,6 +214,11 @@ TextToSpeech::SynthesisResult TextToSpeech::_infer(
     auto* dur_data = dp_outputs[0].GetTensorMutableData<float>();
     std::vector<float> duration(dur_data, dur_data + bsz);
     
+    // Apply speed factor to duration
+    for (auto& dur : duration) {
+        dur /= speed;
+    }
+    
     // Create new tensors for text encoder (previous ones were moved)
     text_ids_tensor = intArrayToTensor(memory_info, text_ids, text_ids_shape);
     text_mask_tensor = arrayToTensor(memory_info, text_mask, text_mask_shape);
@@ -370,6 +376,7 @@ TextToSpeech::SynthesisResult TextToSpeech::call(
     const std::string& text,
     const Style& style,
     int total_step,
+    float speed,
     float silence_duration
 ) {
     if (style.getTtlShape()[0] != 1) {
@@ -381,7 +388,7 @@ TextToSpeech::SynthesisResult TextToSpeech::call(
     float dur_cat = 0.0f;
     
     for (const auto& chunk : text_list) {
-        auto result = _infer(memory_info, {chunk}, style, total_step);
+        auto result = _infer(memory_info, {chunk}, style, total_step, speed);
         
         if (wav_cat.empty()) {
             wav_cat = result.wav;
@@ -406,9 +413,10 @@ TextToSpeech::SynthesisResult TextToSpeech::batch(
     Ort::MemoryInfo& memory_info,
     const std::vector<std::string>& text_list,
     const Style& style,
-    int total_step
+    int total_step,
+    float speed
 ) {
-    return _infer(memory_info, text_list, style, total_step);
+    return _infer(memory_info, text_list, style, total_step, speed);
 }
 
 // ============================================================================

cpp/helper.h

@@ -91,6 +91,7 @@ public:
         const std::string& text,
         const Style& style,
         int total_step,
+        float speed = 1.05f,
         float silence_duration = 0.3f
     );
     
@@ -98,7 +99,8 @@ public:
         Ort::MemoryInfo& memory_info,
         const std::vector<std::string>& text_list,
         const Style& style,
-        int total_step
+        int total_step,
+        float speed = 1.05f
     );
     
     int getSampleRate() const { return sample_rate_; }
@@ -108,7 +110,8 @@ private:
         Ort::MemoryInfo& memory_info,
         const std::vector<std::string>& text_list,
         const Style& style,
-        int total_step
+        int total_step,
+        float speed = 1.05f
     );
     Config cfgs_;
     UnicodeProcessor* text_processor_;