Fangjun Kuang
Committed by GitHub

Add Swift API for KittenTTS (#2476)

... ... @@ -9,6 +9,10 @@ ls -lh
./run-test-version.sh
./run-tts-kitten-en.sh
ls -lh
rm -rf kitten-*
./run-zipformer-ctc-asr.sh
rm -rf sherpa-onnx-zipformer-*
... ...
... ... @@ -20,3 +20,4 @@ decode-file-sense-voice-with-hr
test-version
zipformer-ctc-asr
dolphin-ctc-asr
tts-kitten-en
... ...
... ... @@ -884,13 +884,30 @@ func sherpaOnnxOfflineTtsKokoroModelConfig(
)
}
func sherpaOnnxOfflineTtsKittenModelConfig(
model: String = "",
voices: String = "",
tokens: String = "",
dataDir: String = "",
lengthScale: Float = 1.0
) -> SherpaOnnxOfflineTtsKittenModelConfig {
return SherpaOnnxOfflineTtsKittenModelConfig(
model: toCPointer(model),
voices: toCPointer(voices),
tokens: toCPointer(tokens),
data_dir: toCPointer(dataDir),
length_scale: lengthScale
)
}
func sherpaOnnxOfflineTtsModelConfig(
vits: SherpaOnnxOfflineTtsVitsModelConfig = sherpaOnnxOfflineTtsVitsModelConfig(),
matcha: SherpaOnnxOfflineTtsMatchaModelConfig = sherpaOnnxOfflineTtsMatchaModelConfig(),
kokoro: SherpaOnnxOfflineTtsKokoroModelConfig = sherpaOnnxOfflineTtsKokoroModelConfig(),
numThreads: Int = 1,
debug: Int = 0,
provider: String = "cpu"
provider: String = "cpu",
kitten: SherpaOnnxOfflineTtsKittenModelConfig = sherpaOnnxOfflineTtsKittenModelConfig()
) -> SherpaOnnxOfflineTtsModelConfig {
return SherpaOnnxOfflineTtsModelConfig(
vits: vits,
... ... @@ -898,7 +915,8 @@ func sherpaOnnxOfflineTtsModelConfig(
debug: Int32(debug),
provider: toCPointer(provider),
matcha: matcha,
kokoro: kokoro
kokoro: kokoro,
kitten: kitten
)
}
... ...
#!/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/kitten.html
# to download more models
if [ ! -f ./kitten-nano-en-v0_1-fp16/model.fp16.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/kitten-nano-en-v0_1-fp16.tar.bz2
tar xf kitten-nano-en-v0_1-fp16.tar.bz2
rm kitten-nano-en-v0_1-fp16.tar.bz2
fi
if [ ! -e ./tts-kitten-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-kitten-en.swift ./SherpaOnnx.swift \
-L ../build-swift-macos/install/lib/ \
-l sherpa-onnx \
-l onnxruntime \
-o tts-kitten-en
strip tts-kitten-en
else
echo "./tts-kitten-en exists - skip building"
fi
export DYLD_LIBRARY_PATH=$PWD/../build-swift-macos/install/lib:$DYLD_LIBRARY_PATH
./tts-kitten-en
... ...
class MyClass {
func playSamples(samples: [Float]) {
print("Play \(samples.count) samples")
}
}
func run() {
let model = "./kitten-nano-en-v0_1-fp16/model.fp16.onnx"
let voices = "./kitten-nano-en-v0_1-fp16/voices.bin"
let tokens = "./kitten-nano-en-v0_1-fp16/tokens.txt"
let dataDir = "./kitten-nano-en-v0_1-fp16/espeak-ng-data"
let kitten = sherpaOnnxOfflineTtsKittenModelConfig(
model: model,
voices: voices,
tokens: tokens,
dataDir: dataDir
)
let modelConfig = sherpaOnnxOfflineTtsModelConfig(debug: 0, kitten: kitten)
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 =
"Friends fell out often because life was changing so fast. The easiest thing in the world was to lose touch with someone."
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-kitten-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()
}
}
... ...