Distil-Whisper Small.en (GGUF)
A Q8_0 GGUF conversion of Distil-Whisper Small.en for transcribe.cpp, the runtime behind local transcription in Glimpse. It runs on CPU, Metal and Vulkan.
| File | Size | Notes |
|---|---|---|
distil-small.en-Q8_0.gguf |
189.0 MB | Q8_0 |
ggml-distil-small.en-q8_0.bin |
183.8 MB | whisper.cpp ggml, Q8_0 (earlier Glimpse versions) |
On Apple Silicon, Glimpse can pair it with the Core ML encoder in Glimpse-Dictation/Whisper-Small.en-coreml.
Provenance
Converted from distil-whisper/distil-small.en with scripts/convert-whisper.py from our
transcribe.cpp fork, then quantized to Q8_0. The GGUF metadata lists Apache 2.0;
the upstream model is MIT, which is the license that applies.
ggml-distil-small.en-q8_0.bin is an unmodified copy of Pomni/distil-small.en-ggml-allquants at commit aba251dc.
| File | SHA-256 |
|---|---|
distil-small.en-Q8_0.gguf |
552869c2c9ac97f03da6497e5ffabdb81b2dee50a81b7bcfb67344c41cadeaa6 |
ggml-distil-small.en-q8_0.bin |
8564c3a318d354992fc4654044f48908514783209b4f09ff043ecb8a0c1ebe8e |
Glimpse
This model powers on-device dictation in Glimpse, a free, open-source dictation app for Mac and Windows. The source is on GitHub.
License
The model and its weights come from the Distil-Whisper project, released under the MIT license (LICENSE). This repository only converts the original checkpoint. It is not an official release. See the original model card for training data and intended use.