| license: mit | |
| tags: | |
| - whisper | |
| - speech-to-text | |
| - stt | |
| - ggml | |
| - mirror | |
| library_name: whisper.cpp | |
| # Whisper GGML Mirror (A.I.M.I) | |
| Mirror of [ggerganov/whisper.cpp](https://huggingface.co/ggerganov/whisper.cpp) Large-v3 + Medium GGML weights for A.I.M.I's Speech-to-Text (STT) feature. Contents unmodified. | |
| ## Files | |
| | File | Size | Use | | |
| |---|---|---| | |
| | `ggml-large-v3.bin` | ~2.9 GB | STT quality tier — best transcription, 16 GB+ VRAM | | |
| | `ggml-medium.bin` | ~1.5 GB | STT default — good quality, smaller footprint | | |
| ## License | |
| **MIT** (OpenAI Whisper + whisper.cpp both MIT). Redistributed unchanged. | |
| ## Attribution | |
| - **Whisper**: Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey, Ilya Sutskever (OpenAI) — *Robust Speech Recognition via Large-Scale Weak Supervision* (2022). | |
| - **whisper.cpp GGML conversion**: Georgi Gerganov. | |