Spaces:
Running on Zero
Running on Zero
| title: speakwrite-azerbaijani | |
| emoji: 🎙️ | |
| colorFrom: yellow | |
| colorTo: indigo | |
| sdk: gradio | |
| sdk_version: 6.31.0 | |
| app_file: app.py | |
| short_description: Azerbaijani speech recognition with KenLM decoding | |
| python_version: "3.12" | |
| startup_duration_timeout: 30m | |
| license: cc-by-nc-4.0 | |
| models: | |
| - LocalDoc/speakwrite-azerbaijani | |
| # speakwrite-azerbaijani | |
| Gradio demo of [LocalDoc/speakwrite-azerbaijani](https://huggingface.co/LocalDoc/speakwrite-azerbaijani): Azerbaijani ASR | |
| (FastConformer-CTC + loop refiner + product-key acoustic memory + 6-gram token KenLM). | |
| - Model: **"speakwrite-azerbaijani by LocalDoc"**, CC BY-NC 4.0 — non-commercial use only. | |
| - Inference code: vendored from [balbal](https://github.com/erayyap/balbal) at commit `1be9f88b77c63fbafa7713c6e46e79847f6c390e` | |
| (Apache-2.0, see `balbal/LICENSE` and `balbal/NOTICE`), with the model repo's `refiner.json`. | |
| - Example audio: [FLEURS](https://huggingface.co/datasets/google/fleurs) Azerbaijani test split | |
| (via [LocalDoc/fleurs-azerbaijani-asr](https://huggingface.co/datasets/LocalDoc/fleurs-azerbaijani-asr)), CC BY 4.0, Google. | |