YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
whisper-codeswitch-ggml
GGML weights for Interview Assist — Russian + English code-switching speech recognition.
Source model
- Base fine-tune: coriollon/whisper-large-v3-turbo-russian-codeswitch
- Runtime: whisper.cpp (
whisper-cli) - Conversion:
Scripts/convert-codeswitch-ggml.shin the Interview Assist repo
Files
| File | Description |
|---|---|
ggml-codeswitch-v3-turbo-ru.bin |
F16 GGML model (~1.5 GB) |
License
Apache 2.0 (inherited from OpenAI Whisper and the coriollon fine-tune).
Usage
Download URL used by the app:
https://huggingface.co/utkonoser/whisper-codeswitch-ggml/resolve/main/ggml-codeswitch-v3-turbo-ru.bin
Manual test with whisper-cli:
whisper-cli -m ggml-codeswitch-v3-turbo-ru.bin -f audio.wav -l ru --no-timestamps -otxt
Recommended language flag: ru (model is optimized for Russian speech with embedded English tech terms).
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support