deepfilternet3 / README.md
Samx20's picture
DeepFilterNet 3: the exact files Rostrum downloads, with model card and licence
debc803 verified
|
Raw History Blame Contribute Delete
2.38 kB
---
license: apache-2.0
pipeline_tag: audio-to-audio
library_name: onnx
tags:
- onnx
- rostrum
- audio
- noise-suppression
- speech-enhancement
- deepfilternet
---
# DeepFilterNet 3 (ONNX, torchDF export), mirrored for Rostrum
DeepFilterNet 3, a real-time full-band (48 kHz) speech enhancement model that removes noise from under a voice, as a single-graph ONNX export made with the torchDF export path.
This repository is a pinned, public copy of the exact files the [Rostrum](https://huggingface.co/rostrumstudios) apps download. Rostrum fetches each file at a fixed commit of this repository and checks its SHA-256 before use.
## What Rostrum uses it for
Sound's Clean up and Compose's dialogue: it takes noise (hum, room tone, traffic) out from under a recorded voice. It changes no words and makes no new voice.
Rostrum uses it for **analysis only, never generation**: it reads what is in a recording or a picture and makes no new sound, voice or image.
## Files
| File | Bytes | SHA-256 |
|---|---:|---|
| `denoiser_model.onnx` | 16,104,687 | `fe5eb64fa2e4154c83f8e4935e82871c850c154387ee892e0ab65fe179e7d8c9` |
In Rostrum's models folder they go under `deepfilternet3/`.
## Source
| What | Where | Commit / version |
|---|---|---|
| Exact file (unchanged copy) | https://huggingface.co/kimtos-labs/denoiser-dfn3 | 888f33c41851d7168ae31c3637c323b9e2f3c2a4 |
| Original model, Hendrik Schröter et al. | https://github.com/Rikorose/DeepFilterNet | 7b14b895fe7fc42cab72b4a4c1e49c982bd6404b (main when mirrored) |
| ONNX export path (torchDF) | https://github.com/grazder/DeepFilterNet | 408414f2bdb9ed772580558d1c3ec26c5dc002d7 (torchDF_main when mirrored) |
## Authors
Hendrik Schröter, Tobias Rosenkranz, Alberto N. Escalante-B., Andreas Maier (DeepFilterNet); grazder (torchDF export); kimtos-labs (the hosted export file).
## Licence
DeepFilterNet is dual-licensed **MIT OR Apache-2.0**, at your option (`LICENSE-MIT`, `LICENSE-APACHE`, from the original repository). File redistributed unchanged.
## Citation
The original authors ask that you cite:
```bibtex
@inproceedings{schroeter2023deepfilternet3,
title = {{DeepFilterNet}: Perceptually Motivated Real-Time Speech Enhancement},
author = {Schröter, Hendrik and Rosenkranz, Tobias and Escalante-B., Alberto N. and Maier, Andreas},
booktitle={INTERSPEECH},
year = {2023},
}
```