SpeechT5Checkpoints

Models & Types · AiDotNet.TextToSpeech.MultiModal

The files SpeechT5’s release reads.

Constructors

Constructor Summary
SpeechT5Checkpoints()

Properties

Property Summary
string Model { get; set; } The model: microsoft/speecht5_tts’s pytorch_model.bin (Hugging Face names), or an official fairseq checkpoint (its names, optionally under model.).
string SpeakerEncoder { get; set; } speechbrain’s x-vector extractor, speechbrain/spkrec-xvect-voxceleb’s embedding_model.ckpt; null keeps it as it is.
string Vocoder { get; set; } HiFi-GAN: microsoft/speecht5_hifigan’s pytorch_model.bin, with its spectrogram mean and scale; null keeps the vocoder as it is.