Languages
Whisper is multilingual, and OpenVoiceFlow exposes eleven languages in its picker. Which model you run matters more here than anywhere else in the app.
Languages in the picker
Dashboard ▸ Settings ▸ Transcription — on this Mac ▸ Language offers:
| Language | Code | Language | Code |
|---|---|---|---|
| English | en | Dutch | nl |
| Spanish | es | Hindi | hi |
| French | fr | Japanese | ja |
| German | de | Chinese | zh |
| Italian | it | Korean | ko |
| Portuguese | pt |
Whisper ships both multilingual models and English-only variants (their ids end in .en). The four engines in the onboarding picker — Tiny, Small, Medium, and Large turbo — are multilingual and work with every language above. If your model setting shows an id ending in .en, it will not transcribe other languages properly. Re-pick an engine in Settings ▸ Transcription and the pairing takes care of itself.
Accuracy outside English
Whisper's quality varies by language, and the gap between model sizes widens considerably away from English. As models get smaller they effectively become English specialists. Practical guidance:
- For non-English dictation, run the largest model your Mac handles comfortably before tuning anything else.
- Widely represented languages (Spanish, French, German, Portuguese, Italian) do well. Languages with less training data are more variable.
- The personal dictionary works in any language and is the fix for names regardless.
Switching languages
The language setting is a single global choice, not per-app or automatic. Change it in Settings when you switch languages. There is no mixed-language mode: dictate one language per take for best results.
Cleanup and language
If you use AI cleanup, the model you point it at also needs to handle your language well. Cleanup is instructed to preserve meaning and return only the cleaned text, so a model weak in your language may do more harm than good. When in doubt in a non-English language, leaving cleanup off and taking the raw Whisper transcript is often the better result.
Documentation for OpenVoiceFlow 0.5.5 · last updated 2026-07-31. Something wrong or missing here? That is a bug — tell us at shimoverse@gmail.com.