Vocoder with HiFIGAN Unit trained on LJSpeech

This repository provides all the necessary tools for using a HiFiGAN Unit vocoder trained with LJSpeech.

The pre-trained model take as input discrete self-supervised representations and produces a waveform as output. Typically, this model is utilized on top of a speech-to-unit translation model that converts an input utterance from a source language into a sequence of discrete speech units in a target language.
To generate the discrete self-supervised representations, we employ a K-means clustering model trained on the 6th layer of HuBERT, with k=100.

Install SpeechBrain

First of all, please install tranformers and SpeechBrain with the following command:

pip install speechbrain transformers

Please notice that we encourage you to read our tutorials and learn more about SpeechBrain.

Using the Vocoder

import torch
from speechbrain.inference.vocoders import UnitHIFIGAN

hifi_gan_unit = UnitHIFIGAN.from_hparams(source="speechbrain/tts-hifigan-unit-hubert-l6-k100-ljspeech", savedir="pretrained_models/tts-hifigan-unit-hubert-l6-k100-ljspeech")
codes = torch.randint(0, 99, (100, 1))
waveform = hifi_gan_unit.decode_unit(codes)

Using the Vocoder with the S2UT

import torch
import torchaudio
from speechbrain.inference.ST import EncoderDecoderS2UT
from speechbrain.inference.vocoders import UnitHIFIGAN

# Intialize S2UT (Transformer) and Vocoder (HiFIGAN Unit)
s2ut = EncoderDecoderS2UT.from_hparams(source="speechbrain/s2st-transformer-fr-en-hubert-l6-k100-cvss", savedir="pretrained_models/s2st-transformer-fr-en-hubert-l6-k100-cvss")
hifi_gan_unit = UnitHIFIGAN.from_hparams(source="speechbrain/tts-hifigan-unit-hubert-l6-k100-ljspeech", savedir="pretrained_models/tts-hifigan-unit-hubert-l6-k100-ljspeech")

# Running the S2UT model
codes = s2ut.translate_file("speechbrain/s2st-transformer-fr-en-hubert-l6-k100-cvss/example-fr.wav")
codes = torch.IntTensor(codes[:-1])

# Running Vocoder (units-to-waveform)
waveforms = hifi_gan_unit.decode_unit(codes.unsqueeze(-1))

# Save the waverform
torchaudio.save('example.wav',waveforms.squeeze(1), 16000)

Inference on GPU

To perform inference on the GPU, add run_opts={"device":"cuda"} when calling the from_hparams method.

Limitations

The SpeechBrain team does not provide any warranty on the performance achieved by this model when used on other datasets.

Referencing SpeechBrain

@misc{SB2021,
    author = {Ravanelli, Mirco and Parcollet, Titouan and Rouhe, Aku and Plantinga, Peter and Rastorgueva, Elena and Lugosch, Loren and Dawalatabad, Nauman and Ju-Chieh, Chou and Heba, Abdel and Grondin, Francois and Aris, William and Liao, Chien-Feng and Cornell, Samuele and Yeh, Sung-Lin and Na, Hwidong and Gao, Yan and Fu, Szu-Wei and Subakan, Cem and De Mori, Renato and Bengio, Yoshua },
    title = {SpeechBrain},
    year = {2021},
    publisher = {GitHub},
    journal = {GitHub repository},
    howpublished = {\\\\url{https://github.com/speechbrain/speechbrain}},
  }

About SpeechBrain

SpeechBrain is an open-source and all-in-one speech toolkit. It is designed to be simple, extremely flexible, and user-friendly. Competitive or state-of-the-art performance is obtained in various domains.

Website: https://speechbrain.github.io/

GitHub: https://github.com/speechbrain/speechbrain