https://github.com/thewh1teagle/israwave
Mission to create a Hebrew TTS model as powerful and user-friendly as WaveNet
https://github.com/thewh1teagle/israwave
hebrew israel pytorch tts
Last synced: over 1 year ago
JSON representation
Mission to create a Hebrew TTS model as powerful and user-friendly as WaveNet
- Host: GitHub
- URL: https://github.com/thewh1teagle/israwave
- Owner: thewh1teagle
- Created: 2024-06-21T21:37:05.000Z (about 2 years ago)
- Default Branch: main
- Last Pushed: 2025-01-05T04:24:20.000Z (over 1 year ago)
- Last Synced: 2025-04-10T01:08:25.830Z (over 1 year ago)
- Topics: hebrew, israel, pytorch, tts
- Language: Python
- Homepage: https://huggingface.co/spaces/thewh1teagle/tts-with-israwave
- Size: 178 KB
- Stars: 33
- Watchers: 4
- Forks: 8
- Open Issues: 7
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# israwave
Mission to create a Hebrew TTS model as powerful and user-friendly as WaveNet
## Features
- Generate sentence in less than 1ms on CPU
- Powerful text processor by espeak-ng
- Support for SSML (soon)
## Play with it!
You can play with it on [HuggingFace Space](https://huggingface.co/spaces/thewh1teagle/tts-with-israwave)
## Samples
https://github.com/user-attachments/assets/3212a800-406f-4d79-8aa1-d814eed815d6
## Setup
```console
pip install -U israwave
```
- You also need to place the files [`israwave.onnx`](https://github.com/thewh1teagle/israwave/releases/download/v0.1.0/israwave.onnx), [`nakdimon.onnx`](https://github.com/thewh1teagle/israwave/releases/download/v0.1.0/nakdimon.onnx), and extracted [`espeak-ng-data`](https://github.com/thewh1teagle/israwave/releases/download/v0.1.0/espeak-ng-data.tar.gz) in the project folder.
- We recommend use [uv](https://docs.astral.sh/uv/getting-started/installation) for Python.
- Python 3.12 is recommended. (`uv init -p 3.12`)
- Please see examples.
## Examples
See [examples](examples)
## Dataset
The model trained on [saspeech gold standard](https://openslr.org/134/).
## Thanks
Thanks to [Kan11](https://www.kan.org.il/) and [Shaul](https://www.kan.org.il/authors/%D7%A9%D7%90%D7%95%D7%9C-%D7%90%D7%9E%D7%A1%D7%98%D7%A8%D7%93%D7%9E%D7%A1%D7%A7%D7%99/) for providing the dataset.
Thanks to [elazarg](https://github.com/elazarg) for sharing the [Nakdimon](https://github.com/elazarg/nakdimon) diacritics model, which was instrumental in our project.
For [mush42](https://github.com/mush42) for their excellent TTS training recipe.