File size: 1,852 Bytes
d90bcda | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 | ---
license: other
library_name: vokra
tags:
- vokra
- gguf
---
# deepfilternet3 (Vokra GGUF)
Converted to the Vokra GGUF format for [Vokra](https://github.com/ayutaz/vokra), a zero-dependency speech-AI inference runtime.
**This is a conversion, not a new model.** The weights are the upstream ones; Vokra re-packages them so its runtime can memory-map them directly. Credit for the model belongs upstream — see *Source* below.
## Files
| File | Size | SHA-256 |
|---|---|---|
| `dfn3.gguf` | 8.2 MB | `448c391760a5bc7e6acb63698796dadc959c7439cd2fb56cca59e0bdd7a68b1e` |
## Usage
```bash
# Download (any HTTP client works — the file is a plain GGUF)
curl -L -o dfn3.gguf \
https://huggingface.co/vokra/deepfilternet3/resolve/main/dfn3.gguf
```
```bash
vokra-cli run --model dfn3.gguf --input input.wav
```
## Provenance
| Field | Value |
|---|---|
| Architecture | `denoise` |
| Tensors | 115 |
| Upstream source | Rikorose/DeepFilterNet (MIT OR Apache-2.0) |
| Upstream licence | `MIT OR Apache-2.0` |
| Licence class | `permissive` |
| Registry model id | `deepfilternet3` |
| Vokra GGUF schema | 1 |
| Converted by | vokra-core 0.1.0-alpha.0 |
Every row above is read out of this file's own `vokra.*` metadata, so the card cannot claim something the artifact does not carry.
## Licence
The weights are distributed under **MIT OR Apache-2.0**, unchanged from upstream. Conversion does not alter the licence, and your obligations run to the upstream author.
> The Hugging Face `license:` tag is set to `other` because `MIT OR Apache-2.0` is not one of Hugging Face's accepted tag values. The authoritative licence is the one named above and reproduced in the `LICENSE` file — not the tag.
## Verifying this file
```bash
shasum -a 256 dfn3.gguf
# expect: 448c391760a5bc7e6acb63698796dadc959c7439cd2fb56cca59e0bdd7a68b1e
```
|