Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Unofficial PyTorch Implementation of UnivNet Vocoder (https://arxiv.org/abs/2106.07889)
| Date | Stars |
|---|---|
| 2026-07-24 | 286 |
| 2026-07-25 | 286 |
| 2026-07-28 | 286 |
| 2026-07-30 | 286 |
| 2026-08-06 | 286 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
# UnivNet
**UnivNet: A Neural Vocoder with Multi-Resolution Spectrogram Discriminators for High-Fidelity Waveform Generation**
This is an unofficial PyTorch implementation of ***Jang et al.* (Kakao), [UnivNet](https://arxiv.org/abs/2106.07889)**.
Audio samples are uploaded!
[](https://arxiv.org/abs/2106.07889) [](https://mindslab-ai.github.io/univnet/) [](./LICENSE)
## Notes
**Both UnivNet-c16 and c32 results and the pre-trained weights have been uploaded.**
**For both models, our implementation matches the objective scores (PESQ and RMSE) of the original paper.**
## Key Features
<img src="docs/model_architecture.png" width="100%">
- According to the authors of the paper, UnivNet obtained the best objective results among the recent GAN-based neural vocoders (including HiFi-GAN) as well as outperforming HiFi-GAN in a subjective evaluation. Also its inference speed is 1.5 times faster than HiFi-GAN.
- This repository uses the same mel-spectrogram function as the [Official HiFi-GAN](https://github.com/jik876/hifi-gan), which is compatible with [NVIDIA/tacotron2](https://github.com/NVIDIA/tacotron2).
- Our default mel calculation hyperparameters are as below, following the original paper.
```yaml
audio:
n_mel_channels: 100
filter_length: 1024
hop_length: 256 # WARNING: this can't be changed.
win_length: 1024
sampling_rate: 24000
mel_fmin: 0.0
mel_fmax: 12000.0
```
You can modify the hyperparameters to be compatible with your acoustic model.
## Prerequisites
The implementation needs following dependencies.
0. Python 3.6
1. [PyTorch](https://pytorch.org/) 1.6.0
2. [NumPy](https://numpy.org/) 1.17.4 and [SciPy](https://www.scipy.org/) 1.5.4
3. Install other dependencies in [requirements.txt](./requirements.txt).
```bash
pip install -r requirements.txt
```
## Datasets
**Preparing Data**
- Download the training dataset. This can be any wav file with sampling rate 24,000Hz. The original paper used LibriTTS.
- LibriTTS train-clean-360 split [tar.gz link](https://www.openslr.org/resources/60/train-clean-360.tar.gz)
- Unzip and place its contents under `datasets/LibriTTS/train-clean-360`.
- If you want to use wav files with a different sampling rate, please edit the configuration file (see below).
Note: The mel-spectrograms calculated from audio file will be saved as `**.mel` at first, and then loaded from disk afterwards.
**Preparing Metadata**
Following the format from [NVIDIA/tacotron2](https://github.com/NVIDIA/tacotron2), the metadata should be formatted as:
```
path_to_wav|transcript|speaker_id
path_to_wav|transcript|speaker_id
...
```
Train/validation metadata for LibriTTS train-clean-360 split and are already prepared in `datasets/metadata`.
5% of the train-clean-360 utterances were randomly sampled for validation.
Since this model is a vocoder, the transcripts are **NOT** used during training.
## Train
**Preparing Configuration Files**
- Run `cp config/default_c32.yaml config/config.yaml` and then edit `config.yaml`
- Write down the root path of train/validation in the `data` section. The data loader parses list of files within the path recursively.
```yaml
data:
train_dir: 'datasets/' # root path of train data (either relative/absoulte path is ok)
train_meta: 'metadata/libritts_train_clean_360_train.txt' # relative path of metadata file from train_dir
val_dir: 'datasets/' # root path of validation data
val_meta: 'metadata/libritts_train_clean_360_val.txt' # relative path of metadata file from val_dir
```
We provide the default metadata for LibriTTS train-clean-360 split.
- Modify `channel_size`Excerpt of 7,649 characters
Read on GitHubWould you bet a product on this? Bounded 0–100 and slow moving.
matched fp:bc477bc7d2547a8b, topic:text-to-speech, topic:tts, topic:speech-synthesis
matched fp:bc477bc7d2547a8b, topic:deep-learning, topic:pytorch