loveisgone/super-speech-denoise

Container Template for SoundsRight Subnet Miners

Miners in Bittensor's SoundsRight Subnet must containerize their models before uploading to HuggingFace. This repo serves as a template.

The branches DENOISING_16000HZ and DEREVERBERATION_16000HZ contain this template fitted with SGMSE+ and are also helpful resources for how to incorporate your model.

The main branch contains a template for a container that will spin up an API to communicate with the validator. The following entrypoints cannot be altered:

1. /status/ : Communicates API status
2. /prepare/ : Makes necessary preparations (downloading checkpoints, etc.) and initializes model
3. /upload-audio/ : Upload audio files, save to noisy audio directory
4. /enhance/ : Initialize model, enhance audio files, save to enhanced audio directory
5. /download-enhanced/ : Download enhanced audio files

To add your own model to this template, there are a few things that a miner must do:

1. Add the model files under the model directory.
2. Modify the modelapi.prepare method in app/app.py with necessary preparations to initialize your model.
3. Modify the modelapi.enhance method in app/app.py with the logic your model uses to enhance audio.
4. Update dependencies in pyproject.toml with the dependencies used by your model.
5. If you have directories other than app in your repository, be sure to modify the Dockerfile accordingly (reference line 12 in the Dockerfile for how to do this).
6. Cite your sources (if applicable).

For your model to be processed by validators, there are a few formatting requirements. Note that the template already has been formatted to fit these guidelines.

1. API endpoints must as outlined above.
2. Port must be 6500.
3. Container must be configured to run as non-root user.