Downloads · 30 days
25
29% of all-time downloads
vadimbelsky/emirati-hifigan-bilingual-v1.0
emirati-hifigan-bilingual-v1.0 is a machine learning model from vadimbelsky. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for nemo.
Fine-tuned HiFi-GAN vocoder for Emirati Arabic TTS with bilingual support (AR/EN).
Downloads · 30 days
25
29% of all-time downloads
All-time downloads
85
Public
Repo size
790 MB
Likes
0
Public
Click a slice to open those files.
.pt451 MB · 57%
From the Hugging Face model README
Fine-tuned HiFi-GAN vocoder for Emirati Arabic TTS with bilingual support (AR/EN).
Trained to synthesize high-quality audio from mel-spectrograms generated by FastPitch.
from nemo.collections.tts.models import HifiGanModel
import torch
# Load vocoder
vocoder = HifiGanModel.from_pretrained("vadimbelsky/emirati-hifigan-bilingual-v1.0")
vocoder = vocoder.eval().cuda()
# Generate audio from mel-spectrogram
# spec shape: (1, 80, time_steps)
with torch.no_grad():
audio = vocoder.convert_spectrogram_to_audio(spec=mel_spec)
@article{kong2020hifigan,
title={HiFi-GAN: Generative Adversarial Networks for Efficient and High Fidelity Speech Synthesis},
author={Kong, Jungil and Kim, Jaehyun and Bae, Jaesung},
journal={arXiv preprint arXiv:2010.05646},
year={2020}
}
CC-BY-4.0