Add JavaScript (WebAssembly) API for Kokoro TTS models. (#1726)

k2-fsa · Jan 17, 2025 · 3a1de0b · 3a1de0b
1 parent e8d499d
commit 3a1de0b
Show file tree

Hide file tree

Showing 5 changed files with 154 additions and 6 deletions.
diff --git a/.github/scripts/test-nodejs-npm.sh b/.github/scripts/test-nodejs-npm.sh
@@ -10,7 +10,15 @@ ls -lh
 ls -lh node_modules
 
 # offline tts
-#
+
+curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/kokoro-en-v0_19.tar.bz2
+tar xf kokoro-en-v0_19.tar.bz2
+rm kokoro-en-v0_19.tar.bz2
+
+node ./test-offline-tts-kokoro-en.js
+
+ls -lh
+
 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

diff --git a/nodejs-examples/README.md b/nodejs-examples/README.md
@@ -42,6 +42,22 @@ node ./test-offline-speaker-diarization.js
 
 In the following, we demonstrate how to run text-to-speech.
 
+## ./test-offline-tts-kokoro-en.js
+
+[./test-offline-tts-kokoro-en.js](./test-offline-tts-kokoro-en.js) shows how to use
+[kokoro-en-v0_19](https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/kokoro-en-v0_19.tar.bz2)
+for text-to-speech.
+
+You can use the following command to run it:
+
+```bash
+wget https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/kokoro-en-v0_19.tar.bz2
+tar xf kokoro-en-v0_19.tar.bz2
+rm kokoro-en-v0_19.tar.bz2
+
+node ./test-offline-tts-kokoro-en.js
+```
+
 ## ./test-offline-tts-matcha-zh.js
 
 [./test-offline-tts-matcha-zh.js](./test-offline-tts-matcha-zh.js) shows how to use

diff --git a/nodejs-examples/test-offline-tts-kokoro-en.js b/nodejs-examples/test-offline-tts-kokoro-en.js
@@ -0,0 +1,37 @@
+// Copyright (c)  2025  Xiaomi Corporation (authors: Fangjun Kuang)
+
+const sherpa_onnx = require('sherpa-onnx');
+
+function createOfflineTts() {
+  let offlineTtsKokoroModelConfig = {
+    model: './kokoro-en-v0_19/model.onnx',
+    voices: './kokoro-en-v0_19/voices.bin',
+    tokens: './kokoro-en-v0_19/tokens.txt',
+    dataDir: './kokoro-en-v0_19/espeak-ng-data',
+    lengthScale: 1.0,
+  };
+  let offlineTtsModelConfig = {
+    offlineTtsKokoroModelConfig: offlineTtsKokoroModelConfig,
+    numThreads: 1,
+    debug: 1,
+    provider: 'cpu',
+  };
+
+  let offlineTtsConfig = {
+    offlineTtsModelConfig: offlineTtsModelConfig,
+    maxNumSentences: 1,
+  };
+
+  return sherpa_onnx.createOfflineTts(offlineTtsConfig);
+}
+
+const tts = createOfflineTts();
+const speakerId = 0;
+const speed = 1.0;
+const 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.'
+
+const audio = tts.generate({text: text, sid: speakerId, speed: speed});
+tts.save('./test-kokoro-en.wav', audio);
+console.log('Saved to test-kokoro-en.wav successfully.');
+tts.free();
diff --git a/wasm/tts/sherpa-onnx-tts.js b/wasm/tts/sherpa-onnx-tts.js
@@ -8,8 +8,12 @@ function freeConfig(config, Module) {
     freeConfig(config.config, Module)
   }
 
-  if ('config2' in config) {
-    freeConfig(config.config2, Module)
+  if ('matcha' in config) {
+    freeConfig(config.matcha, Module)
+  }
+
+  if ('kokoro' in config) {
+    freeConfig(config.kokoro, Module)
   }
 
   Module._free(config.ptr);
@@ -132,6 +136,52 @@ function initSherpaOnnxOfflineTtsMatchaModelConfig(config, Module) {
   }
 }
 
+function initSherpaOnnxOfflineTtsKokoroModelConfig(config, Module) {
+  const modelLen = Module.lengthBytesUTF8(config.model) + 1;
+  const voicesLen = Module.lengthBytesUTF8(config.voices) + 1;
+  const tokensLen = Module.lengthBytesUTF8(config.tokens || '') + 1;
+  const dataDirLen = Module.lengthBytesUTF8(config.dataDir || '') + 1;
+
+  const n = modelLen + voicesLen + tokensLen + dataDirLen;
+
+  const buffer = Module._malloc(n);
+
+  const len = 5 * 4;
+  const ptr = Module._malloc(len);
+
+  let offset = 0;
+  Module.stringToUTF8(config.model || '', buffer + offset, modelLen);
+  offset += modelLen;
+
+  Module.stringToUTF8(config.voices || '', buffer + offset, voicesLen);
+  offset += voicesLen;
+
+  Module.stringToUTF8(config.tokens || '', buffer + offset, tokensLen);
+  offset += tokensLen;
+
+  Module.stringToUTF8(config.dataDir || '', buffer + offset, dataDirLen);
+  offset += dataDirLen;
+
+  offset = 0;
+  Module.setValue(ptr, buffer + offset, 'i8*');
+  offset += modelLen;
+
+  Module.setValue(ptr + 4, buffer + offset, 'i8*');
+  offset += voicesLen;
+
+  Module.setValue(ptr + 8, buffer + offset, 'i8*');
+  offset += tokensLen;
+
+  Module.setValue(ptr + 12, buffer + offset, 'i8*');
+  offset += dataDirLen;
+
+  Module.setValue(ptr + 16, config.lengthScale || 1.0, 'float');
+
+  return {
+    buffer: buffer, ptr: ptr, len: len,
+  }
+}
+
 function initSherpaOnnxOfflineTtsModelConfig(config, Module) {
   if (!('offlineTtsVitsModelConfig' in config)) {
     config.offlineTtsVitsModelConfig = {
@@ -159,14 +209,29 @@ function initSherpaOnnxOfflineTtsModelConfig(config, Module) {
     };
   }
 
+  if (!('offlineTtsKokoroModelConfig' in config)) {
+    config.offlineTtsKokoroModelConfig = {
+      model: '',
+      voices: '',
+      tokens: '',
+      lengthScale: 1.0,
+      dataDir: '',
+    };
+  }
+
 
   const vitsModelConfig = initSherpaOnnxOfflineTtsVitsModelConfig(
       config.offlineTtsVitsModelConfig, Module);
 
   const matchaModelConfig = initSherpaOnnxOfflineTtsMatchaModelConfig(
       config.offlineTtsMatchaModelConfig, Module);
 
-  const len = vitsModelConfig.len + matchaModelConfig.len + 3 * 4;
+  const kokoroModelConfig = initSherpaOnnxOfflineTtsKokoroModelConfig(
+      config.offlineTtsKokoroModelConfig, Module);
+
+  const len = vitsModelConfig.len + matchaModelConfig.len +
+      kokoroModelConfig.len + 3 * 4;
+
   const ptr = Module._malloc(len);
 
   let offset = 0;
@@ -188,9 +253,12 @@ function initSherpaOnnxOfflineTtsModelConfig(config, Module) {
   Module._CopyHeap(matchaModelConfig.ptr, matchaModelConfig.len, ptr + offset);
   offset += matchaModelConfig.len;
 
+  Module._CopyHeap(kokoroModelConfig.ptr, kokoroModelConfig.len, ptr + offset);
+  offset += kokoroModelConfig.len;
+
   return {
     buffer: buffer, ptr: ptr, len: len, config: vitsModelConfig,
-        config2: matchaModelConfig
+        matcha: matchaModelConfig, kokoro: kokoroModelConfig,
   }
 }
 
@@ -308,9 +376,18 @@ function createOfflineTts(Module, myConfig) {
     lengthScale: 1.0,
   };
 
+  const offlineTtsKokoroModelConfig = {
+    model: '',
+    voices: '',
+    tokens: '',
+    dataDir: '',
+    lengthScale: 1.0,
+  };
+
   const offlineTtsModelConfig = {
     offlineTtsVitsModelConfig: offlineTtsVitsModelConfig,
     offlineTtsMatchaModelConfig: offlineTtsMatchaModelConfig,
+    offlineTtsKokoroModelConfig: offlineTtsKokoroModelConfig,
     numThreads: 1,
     debug: 1,
     provider: 'cpu',

diff --git a/wasm/tts/sherpa-onnx-wasm-main-tts.cc b/wasm/tts/sherpa-onnx-wasm-main-tts.cc
@@ -15,9 +15,11 @@ extern "C" {
 
 static_assert(sizeof(SherpaOnnxOfflineTtsVitsModelConfig) == 8 * 4, "");
 static_assert(sizeof(SherpaOnnxOfflineTtsMatchaModelConfig) == 8 * 4, "");
+static_assert(sizeof(SherpaOnnxOfflineTtsKokoroModelConfig) == 5 * 4, "");
 static_assert(sizeof(SherpaOnnxOfflineTtsModelConfig) ==
                   sizeof(SherpaOnnxOfflineTtsVitsModelConfig) +
-                      sizeof(SherpaOnnxOfflineTtsMatchaModelConfig) + 3 * 4,
+                      sizeof(SherpaOnnxOfflineTtsMatchaModelConfig) +
+                      sizeof(SherpaOnnxOfflineTtsKokoroModelConfig) + 3 * 4,
               "");
 static_assert(sizeof(SherpaOnnxOfflineTtsConfig) ==
                   sizeof(SherpaOnnxOfflineTtsModelConfig) + 3 * 4,
@@ -27,6 +29,7 @@ void MyPrint(SherpaOnnxOfflineTtsConfig *tts_config) {
   auto tts_model_config = &tts_config->model;
   auto vits_model_config = &tts_model_config->vits;
   auto matcha_model_config = &tts_model_config->matcha;
+  auto kokoro = &tts_model_config->kokoro;
   fprintf(stdout, "----------vits model config----------\n");
   fprintf(stdout, "model: %s\n", vits_model_config->model);
   fprintf(stdout, "lexicon: %s\n", vits_model_config->lexicon);
@@ -47,6 +50,13 @@ void MyPrint(SherpaOnnxOfflineTtsConfig *tts_config) {
   fprintf(stdout, "length scale: %.3f\n", matcha_model_config->length_scale);
   fprintf(stdout, "dict_dir: %s\n", matcha_model_config->dict_dir);
 
+  fprintf(stdout, "----------kokoro model config----------\n");
+  fprintf(stdout, "model: %s\n", kokoro->model);
+  fprintf(stdout, "voices: %s\n", kokoro->voices);
+  fprintf(stdout, "tokens: %s\n", kokoro->tokens);
+  fprintf(stdout, "data_dir: %s\n", kokoro->data_dir);
+  fprintf(stdout, "length scale: %.3f\n", kokoro->length_scale);
+
   fprintf(stdout, "----------tts model config----------\n");
   fprintf(stdout, "num threads: %d\n", tts_model_config->num_threads);
   fprintf(stdout, "debug: %d\n", tts_model_config->debug);