| --- |
| license: mit |
| language: |
| - multilingual |
| library_name: whisper.cpp |
| pipeline_tag: automatic-speech-recognition |
| tags: |
| - whisper |
| - ggml |
| - on-device |
| - offline |
| --- |
| |
| # OfflineOCR Whisper small q5_1 |
| |
| This repository mirrors the multilingual OpenAI Whisper `small` model converted |
| and quantized for `whisper.cpp`. |
| |
| - File: `ggml-small-q5_1.bin` |
| - Size: `190085487` bytes |
| - SHA-256: `ae85e4a935d7a567bd102fe55afc16bb595bdb618e11b2fc7591bc08120411bb` |
| - Runtime: `whisper.cpp` v1.9.1 |
| - Quantization: `q5_1` |
|
|
| OfflineOCR uses Hugging Face as the primary application download and ModelScope |
| as the sequential fallback. Android receives the same verified bytes through |
| the Google Play `whisper_models` on-demand asset pack. |
|
|
| The upstream model weights and OpenAI Whisper code are MIT licensed. The |
| `whisper.cpp` runtime is also MIT licensed. Retain both license notices when |
| redistributing this artifact with an application. |
|
|
