How to get wavfrom from the predicted mels

auspicious3000 / autovc

AutoVC: Zero-Shot Voice Style Transfer with Only Autoencoder Loss

https://arxiv.org/abs/1905.05879

MIT License

976 stars 207 forks source link

Closed JohnHerry closed 3 years ago

JohnHerry commented 3 years ago

Hi, all Did anybody try to generate wavfrom from the autovc generated mels? The wavenet_vocoder is too slow, So we tried to to use librosa.

librosa.feature.inverse.mel_to_audio(mels, sr=16000, n_fft=1024, hop_length=256)

but the result wav sounds like a noise.

need suggestion!