Add Swift API for Kokoro TTS 1.0 (#1803)

.github/scripts/test-swift.sh

@@ -11,6 +11,10 @@ ls -lh
 ls -lh
 rm -rf vits-piper-*
 
+./run-tts-kokoro-zh-en.sh
+ls -lh
+rm -rf kokoro-multi-*
+
 ./run-tts-kokoro-en.sh
 ls -lh
 rm -rf kokoro-en-*

cxx-api-examples/kokoro-tts-en-cxx-api.cc

@@ -3,7 +3,7 @@
 // Copyright (c)  2025  Xiaomi Corporation
 
 // This file shows how to use sherpa-onnx CXX API
-// for Chinese TTS with Kokoro.
+// for English TTS with Kokoro.
 //
 // clang-format off
 /*

cxx-api-examples/kokoro-tts-zh-en-cxx-api.cc

@@ -3,7 +3,7 @@
 // Copyright (c)  2025  Xiaomi Corporation
 
 // This file shows how to use sherpa-onnx CXX API
-// for Chinese TTS with Kokoro.
+// for Chinese + English TTS with Kokoro.
 //
 // clang-format off
 /*

swift-api-examples/.gitignore

@@ -13,3 +13,4 @@ add-punctuations
 tts-matcha-zh
 tts-matcha-en
 tts-kokoro-en
+tts-kokoro-zh-en

swift-api-examples/SherpaOnnx.swift

@@ -767,14 +767,18 @@ func sherpaOnnxOfflineTtsKokoroModelConfig(
   voices: String = "",
   tokens: String = "",
   dataDir: String = "",
-  lengthScale: Float = 1.0
+  lengthScale: Float = 1.0,
+  dictDir: String = "",
+  lexicon: String = ""
 ) -> SherpaOnnxOfflineTtsKokoroModelConfig {
   return SherpaOnnxOfflineTtsKokoroModelConfig(
     model: toCPointer(model),
     voices: toCPointer(voices),
     tokens: toCPointer(tokens),
     data_dir: toCPointer(dataDir),
-    length_scale: lengthScale
+    length_scale: lengthScale,
+    dict_dir: toCPointer(dictDir),
+    lexicon: toCPointer(lexicon)
   )
 }
 

swift-api-examples/run-tts-kokoro-zh-en.sh

@@ -0,0 +1,37 @@
+#!/usr/bin/env bash
+
+set -ex
+
+if [ ! -d ../build-swift-macos ]; then
+  echo "Please run ../build-swift-macos.sh first!"
+  exit 1
+fi
+
+# please visit
+# https://k2-fsa.github.io/sherpa/onnx/tts/pretrained_models/kokoro.html
+# to download more models
+if [ ! -f ./kokoro-multi-lang-v1_0/model.onnx ]; then
+  curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/kokoro-multi-lang-v1_0.tar.bz2
+  tar xf kokoro-multi-lang-v1_0.tar.bz2
+  rm kokoro-multi-lang-v1_0.tar.bz2
+fi
+
+if [ ! -e ./tts-kokoro-zh-en ]; then
+  # Note: We use -lc++ to link against libc++ instead of libstdc++
+  swiftc \
+    -lc++ \
+    -I ../build-swift-macos/install/include \
+    -import-objc-header ./SherpaOnnx-Bridging-Header.h \
+    ./tts-kokoro-zh-en.swift  ./SherpaOnnx.swift \
+    -L ../build-swift-macos/install/lib/ \
+    -l sherpa-onnx \
+    -l onnxruntime \
+    -o tts-kokoro-zh-en
+
+  strip tts-kokoro-zh-en
+else
+  echo "./tts-kokoro-zh-en exists - skip building"
+fi
+
+export DYLD_LIBRARY_PATH=$PWD/../build-swift-macos/install/lib:$DYLD_LIBRARY_PATH
+./tts-kokoro-zh-en

swift-api-examples/tts-kokoro-zh-en.swift

@@ -0,0 +1,69 @@
+class MyClass {
+  func playSamples(samples: [Float]) {
+    print("Play \(samples.count) samples")
+  }
+}
+
+func run() {
+  let model = "./kokoro-multi-lang-v1_0/model.onnx"
+  let voices = "./kokoro-multi-lang-v1_0/voices.bin"
+  let tokens = "./kokoro-multi-lang-v1_0/tokens.txt"
+  let dataDir = "./kokoro-multi-lang-v1_0/espeak-ng-data"
+  let dictDir = "./kokoro-multi-lang-v1_0/dict"
+  let lexicon = "./kokoro-multi-lang-v1_0/lexicon-us-en.txt,./kokoro-multi-lang-v1_0/lexicon-zh.txt"
+  let kokoro = sherpaOnnxOfflineTtsKokoroModelConfig(
+    model: model,
+    voices: voices,
+    tokens: tokens,
+    dataDir: dataDir,
+    dictDir: dictDir,
+    lexicon: lexicon
+  )
+  let modelConfig = sherpaOnnxOfflineTtsModelConfig(kokoro: kokoro, debug: 0)
+  var ttsConfig = sherpaOnnxOfflineTtsConfig(model: modelConfig)
+
+  let myClass = MyClass()
+
+  // We use Unretained here so myClass must be kept alive as the callback is invoked
+  //
+  // See also
+  // https://medium.com/codex/swift-c-callback-interoperability-6d57da6c8ee6
+  let arg = Unmanaged<MyClass>.passUnretained(myClass).toOpaque()
+
+  let callback: TtsCallbackWithArg = { samples, n, arg in
+    let o = Unmanaged<MyClass>.fromOpaque(arg!).takeUnretainedValue()
+    var savedSamples: [Float] = []
+    for index in 0..<n {
+      savedSamples.append(samples![Int(index)])
+    }
+
+    o.playSamples(samples: savedSamples)
+
+    // return 1 so that it continues generating
+    return 1
+  }
+
+  let tts = SherpaOnnxOfflineTtsWrapper(config: &ttsConfig)
+
+  let text =
+    "中英文语音合成测试。This is generated by next generation Kaldi using Kokoro without Misaki. 你觉得中英文说的如何呢？"
+  let sid = 0
+  let speed: Float = 1.0
+
+  let audio = tts.generateWithCallbackWithArg(
+    text: text, callback: callback, arg: arg, sid: sid, speed: speed)
+  let filename = "test-kokoro-zh-en.wav"
+  let ok = audio.save(filename: filename)
+  if ok == 1 {
+    print("\nSaved to:\(filename)")
+  } else {
+    print("Failed to save to \(filename)")
+  }
+}
+
+@main
+struct App {
+  static func main() {
+    run()
+  }
+}