Refresh README — uniform WindyWord template with WER tier + dialect notes
Browse files
README.md
CHANGED
|
@@ -4,34 +4,38 @@ tags:
|
|
| 4 |
- automatic-speech-recognition
|
| 5 |
- whisper
|
| 6 |
- windyword
|
| 7 |
-
- ctranslate2
|
| 8 |
- chinese
|
| 9 |
- chinese
|
| 10 |
-
library_name:
|
| 11 |
pipeline_tag: automatic-speech-recognition
|
| 12 |
language:
|
| 13 |
- zh
|
| 14 |
---
|
| 15 |
|
| 16 |
-
# WindyWord.ai STT — Chinese (Mandarin) Lingua (CPU INT8
|
| 17 |
|
| 18 |
**Transcribes Chinese (Mandarin) speech (Sino-Tibetan > Sinitic).**
|
| 19 |
|
| 20 |
-
|
| 21 |
|
| 22 |
-
|
|
|
|
|
|
|
|
|
|
| 23 |
|
| 24 |
-
|
| 25 |
-
import ctranslate2
|
| 26 |
-
from huggingface_hub import snapshot_download
|
| 27 |
|
| 28 |
-
|
| 29 |
-
model = ctranslate2.models.Whisper(f"{local}/ct2-int8/")
|
| 30 |
-
```
|
| 31 |
|
| 32 |
-
|
| 33 |
|
| 34 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 35 |
|
| 36 |
## Commercial Use
|
| 37 |
|
|
@@ -41,6 +45,6 @@ Visit [windyword.ai](https://windyword.ai) for apps and API access.
|
|
| 41 |
|
| 42 |
## Provenance & License
|
| 43 |
|
| 44 |
-
|
| 45 |
|
| 46 |
*Certified by Opus 4.6 Opus-Claw (Dr. C) on Veron-1 (RTX 5090, Mt Pleasant SC).*
|
|
|
|
| 4 |
- automatic-speech-recognition
|
| 5 |
- whisper
|
| 6 |
- windyword
|
|
|
|
| 7 |
- chinese
|
| 8 |
- chinese
|
| 9 |
+
library_name: transformers
|
| 10 |
pipeline_tag: automatic-speech-recognition
|
| 11 |
language:
|
| 12 |
- zh
|
| 13 |
---
|
| 14 |
|
| 15 |
+
# WindyWord.ai STT — Chinese (Mandarin) Lingua (CPU INT8 (CTranslate2))
|
| 16 |
|
| 17 |
**Transcribes Chinese (Mandarin) speech (Sino-Tibetan > Sinitic).**
|
| 18 |
|
| 19 |
+
## Quality
|
| 20 |
|
| 21 |
+
- **FLEURS WER:** 0.0% (50-sample audit)
|
| 22 |
+
- **CER:** 0.0
|
| 23 |
+
- **Tier:** EXCELLENT ⭐⭐⭐⭐⭐
|
| 24 |
+
- **Source:** WindyWord Grand Rounds v2 audit (50-sample FLEURS)
|
| 25 |
|
| 26 |
+
## About this variant
|
|
|
|
|
|
|
| 27 |
|
| 28 |
+
This is the **ct2-int8** deployment format of our Chinese (Mandarin) Lingua STT model. Load it via the `ct2-int8/` subfolder.
|
|
|
|
|
|
|
| 29 |
|
| 30 |
+
Part of the [WindyWord.ai](https://windyword.ai) STT fleet — covering 35+ languages that commercial speech-to-text APIs underserve, with proper dialect / script disclosures where they matter.
|
| 31 |
|
| 32 |
+
## Usage
|
| 33 |
+
|
| 34 |
+
```python
|
| 35 |
+
from transformers import WhisperForConditionalGeneration, WhisperProcessor
|
| 36 |
+
processor = WhisperProcessor.from_pretrained("WindyWord/listen-windy-lingua-chinese-ct2", subfolder="ct2-int8")
|
| 37 |
+
model = WhisperForConditionalGeneration.from_pretrained("WindyWord/listen-windy-lingua-chinese-ct2", subfolder="ct2-int8")
|
| 38 |
+
```
|
| 39 |
|
| 40 |
## Commercial Use
|
| 41 |
|
|
|
|
| 45 |
|
| 46 |
## Provenance & License
|
| 47 |
|
| 48 |
+
Weights derived from upstream community Whisper fine-tunes (see individual model card for exact lineage). Redistributed under Apache-2.0 (inherited).
|
| 49 |
|
| 50 |
*Certified by Opus 4.6 Opus-Claw (Dr. C) on Veron-1 (RTX 5090, Mt Pleasant SC).*
|