https://github.com/philipamadasun/whisper_torch
An implmenetation of whisper "turbo" model in torch
https://github.com/philipamadasun/whisper_torch
deep-learning fine-tuning torch torchaudio whisper
Last synced: 3 months ago
JSON representation
An implmenetation of whisper "turbo" model in torch
- Host: GitHub
- URL: https://github.com/philipamadasun/whisper_torch
- Owner: PhilipAmadasun
- Created: 2025-03-29T19:20:28.000Z (3 months ago)
- Default Branch: main
- Last Pushed: 2025-03-29T19:53:42.000Z (3 months ago)
- Last Synced: 2025-03-29T20:25:18.514Z (3 months ago)
- Topics: deep-learning, fine-tuning, torch, torchaudio, whisper
- Language: Python
- Homepage:
- Size: 744 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# whisper_torch
An implementation of whisper `turbo` model in torch (class and feed forward method). Uses the pretrained weights for the `turbo` model.
### Library Versions
The following library versions were used for this work:
- torch 2.6.0
- torchaudio 2.6.0
- transformers 4.47.1
To test and compare inference with original whisper implementation:
```
python3 test.py
```
Torch model currently produces wildy inaccurate inference on shorter audio clips as beam search and other optimizations are not yet implemented.