https://github.com/lucadellalib/bigvgan
A single-file implementation of BigVGAN generator
https://github.com/lucadellalib/bigvgan
audio-generation audio-synthesis bigvgan music-synthesis neural-vocoder pytorch singing-voice-synthesis speech-synthesis
Last synced: 28 days ago
JSON representation
A single-file implementation of BigVGAN generator
- Host: GitHub
- URL: https://github.com/lucadellalib/bigvgan
- Owner: lucadellalib
- License: apache-2.0
- Created: 2024-09-08T00:38:52.000Z (8 months ago)
- Default Branch: master
- Last Pushed: 2024-09-08T01:08:11.000Z (8 months ago)
- Last Synced: 2024-10-06T02:06:05.590Z (7 months ago)
- Topics: audio-generation, audio-synthesis, bigvgan, music-synthesis, neural-vocoder, pytorch, singing-voice-synthesis, speech-synthesis
- Language: Python
- Homepage:
- Size: 395 KB
- Stars: 1
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# BigVGAN
A single-file implementation of BigVGAN generator (see https://arxiv.org/abs/2206.04658).
The original implementation can be found at https://github.com/NVIDIA/BigVGAN/tree/v2.4.---------------------------------------------------------------------------------------------------------
## 🛠️️ Installation
First of all, install [Python 3.8 or later](https://www.python.org).
Clone or download and extract the repository, open a terminal and run:```
pip install -r requirements.txt
```---------------------------------------------------------------------------------------------------------
## ▶️ Quickstart
```python
import torch
import torchaudio
from bigvgan import BigVGANdevice = "cuda" if torch.cuda.is_available() else "cpu"
model = BigVGAN().to(device)
model.eval().requires_grad_(False)
wav_path = "sample.wav"
sig, orig_sample_rate = torchaudio.load(wav_path)
sig = sig.to(device)
with torch.no_grad():
feats = model.extract_features(sig, orig_sample_rate)
rec_sig = model(feats).cpu()
torchaudio.save("reconstruction.wav", rec_sig[0, :], model.sample_rate)
```---------------------------------------------------------------------------------------------------------
## 📧 Contact
[[email protected]](mailto:[email protected])
---------------------------------------------------------------------------------------------------------