Add SenseVoice int8 (byte-for-byte redistribution of upstream 2024-07-17 export)
Browse files- README.md +61 -0
- sense-voice/model.int8.onnx +3 -0
- sense-voice/tokens.txt +0 -0
README.md
ADDED
|
@@ -0,0 +1,61 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
language:
|
| 4 |
+
- zh
|
| 5 |
+
- en
|
| 6 |
+
- ja
|
| 7 |
+
- ko
|
| 8 |
+
- yue
|
| 9 |
+
tags:
|
| 10 |
+
- automatic-speech-recognition
|
| 11 |
+
- onnx
|
| 12 |
+
- sherpa-onnx
|
| 13 |
+
- redistribution
|
| 14 |
+
---
|
| 15 |
+
|
| 16 |
+
# praxis-models
|
| 17 |
+
|
| 18 |
+
Model assets redistributed for the **Praxis** language-learning app.
|
| 19 |
+
|
| 20 |
+
**This repository contains no original work.** Every file here is a byte-for-byte
|
| 21 |
+
copy of an upstream release, re-hosted only so the app has a stable, version-pinned
|
| 22 |
+
download source. Nothing has been re-trained, re-quantized, or otherwise modified.
|
| 23 |
+
|
| 24 |
+
## Why re-host at all?
|
| 25 |
+
|
| 26 |
+
The app downloads models on demand and verifies them with a SHA-256 that is
|
| 27 |
+
compiled into the client. Pointing directly at an upstream repository would mean
|
| 28 |
+
that any upstream change — a re-export, a file rename, a repository removal —
|
| 29 |
+
silently breaks downloads for every already-shipped client. Re-hosting gives us a
|
| 30 |
+
path we control; upstream is kept as a fallback mirror.
|
| 31 |
+
|
| 32 |
+
## Contents
|
| 33 |
+
|
| 34 |
+
### `sense-voice/` — SenseVoice ASR (int8)
|
| 35 |
+
|
| 36 |
+
| File | Size (bytes) | SHA-256 |
|
| 37 |
+
|---|---|---|
|
| 38 |
+
| `model.int8.onnx` | 239,233,841 | `c71f0ce00bec95b07744e116345e33d8cbbe08cef896382cf907bf4b51a2cd51` |
|
| 39 |
+
| `tokens.txt` | 315,894 | `f449eb28dc567533d7fa59be34e2abca8784f771850c78a47fb731a31429a1dc` |
|
| 40 |
+
|
| 41 |
+
**Source:** [`csukuangfj/sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17`](https://huggingface.co/csukuangfj/sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17)
|
| 42 |
+
— ONNX export of [SenseVoice](https://github.com/FunAudioLLM/SenseVoice) by FunAudioLLM,
|
| 43 |
+
exported for [sherpa-onnx](https://github.com/k2-fsa/sherpa-onnx) by Fangjun Kuang.
|
| 44 |
+
|
| 45 |
+
The SHA-256 of `model.int8.onnx` matches the upstream LFS pointer exactly, which is
|
| 46 |
+
the check we use to confirm these are unmodified copies.
|
| 47 |
+
|
| 48 |
+
Languages: Chinese, English, Japanese, Korean, Cantonese.
|
| 49 |
+
|
| 50 |
+
## Verifying
|
| 51 |
+
|
| 52 |
+
```bash
|
| 53 |
+
sha256sum model.int8.onnx tokens.txt
|
| 54 |
+
```
|
| 55 |
+
|
| 56 |
+
Any mismatch means the file was corrupted in transit or tampered with — do not use it.
|
| 57 |
+
|
| 58 |
+
## License
|
| 59 |
+
|
| 60 |
+
Apache-2.0, inherited from the upstream releases. All credit for these models
|
| 61 |
+
belongs to their original authors, cited above.
|
sense-voice/model.int8.onnx
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c71f0ce00bec95b07744e116345e33d8cbbe08cef896382cf907bf4b51a2cd51
|
| 3 |
+
size 239233841
|
sense-voice/tokens.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|