A simple C# wrapper of piper TTS application. Similar to piper you can checkout models here and models from huggingface, checkout PiperDownloader usage.
See examples for example usage or tests.
// To download piper executable use
var cwd = Directory.GetCurrentDirectory();
await PiperDownloader.DownloadPiper().ExtractPiper(cwd); // Downloads and extracts piper to cwd/piper directory
...
// You can get list of models from hugging face using
var models = await PiperDownloader.GetHuggingFaceModelList(); // Returns a dictionary with model key as key
var model = models["ar_JO-kareem-low"];
// or you can do
var model = await PiperDownloader.GetModelByKey("ar_JO-kareem-low");
...
// Before you can use the model you need to download it using
model = await model.DownloadModel(cwd);
// Or if its downloaded you can load it from directory
model = await VoiceModel.LoadModel(modelPath);
// Now you can also do
model = await PiperDownloader.DownloadModelByKey("ar_JO-kareem-low");
// Or if its downloaded you can load it by key aswell
model = await VoiceModel.LoadModelByKey("ar_JO-kareem-low");
...
// To start generating audio use PiperProvider
var piperModel = new PiperProvider(new PiperConfiguration()
{
ExecutableLocation = Path.Join(cwd, "piper", "piper.exe"), // Path to piper executable
WorkingDirectory = Path.Join(cwd, "piper"), // Path to piper working directory
Model = model, // Loaded/downloaded VoiceModel
});
// Generate audio, currently supported formats are Mp3, Wav, Raw
var result = await piperModel.InferAsync("Hello there!", AudioOutputType.Wav); // Returns byte[]If you want to report bugs do it here: https://github.com/Lyx52/PiperSharp/issues