Supported meeting transcription languages
Tell Loofah which languages you expect before you record or import audio. The configured languages determine whether Loofah can transcribe the audio and whether the transcript can appear live.
Configure your languages
Section titled “Configure your languages”- Open Settings → App → Language & Region.
- Set Main language to the language you usually speak. It is always included for transcription and is also used for summaries and generated titles.
- Under Additional spoken languages, add every other language that may be spoken in your meetings.
- Open Settings → Transcription and choose a downloaded model that supports the complete selection.
If you do not add a spoken language, Loofah does not use it when checking model support or deciding between live and after-recording transcription. Changing these settings does not re-transcribe existing meetings.
Supported transcription languages
Section titled “Supported transcription languages”The bundled models have different language coverage:
| Model | Transcript timing | Supported languages |
|---|---|---|
| Soniqo Parakeet Streaming | Live | English |
| Soniqo Parakeet Batch | After recording | Bulgarian, Croatian, Czech, Danish, Dutch, English, Estonian, Finnish, French, German, Greek, Hungarian, Italian, Latvian, Lithuanian, Maltese, Polish, Portuguese, Romanian, Russian, Slovak, Slovenian, Spanish, Swedish, and Ukrainian |
| Whisper Base, Small, Large Turbo, and Large V3 (Multilingual) | Live or after recording | Whisper’s multilingual language set, including Dutch and English |
| Whisper Base and Small (English) | Live or after recording | English |
The streaming model is based on NVIDIA’s Parakeet Realtime EOU 120M, which supports only English. The batch model is based on Parakeet TDT 0.6B v3, which supports the 25 European languages listed above.
If Soniqo Parakeet Streaming is selected and every configured language is English, the transcript appears during the meeting. If you configure one or more of the other supported languages, Loofah automatically records first and uses the batch model after you stop.
Language & Region includes more languages than the current transcription models support because the main language also controls generated responses. Every configured spoken language must be supported by your selected model. Settings → Transcription shows a warning when the selected model does not support it.
Choose a Whisper model
Section titled “Choose a Whisper model”Settings → Transcription offers these Whisper variants. Download a model, then select it; the selector shows the active model’s name.
| Model | Approximate download |
|---|---|
| Base (Multilingual or English) | 82 MB |
| Small (Multilingual or English) | 264 MB |
| Large Turbo (Multilingual) | 874 MB |
| Large V3 (Multilingual) | 3.1 GB |
Base, Small, and Large Turbo use 8-bit quantized weights to reduce storage. Choose a multilingual variant if you expect more than English. All Whisper variants support While recording and After recording under Transcription timing. Live text arrives in speech chunks. Imported audio is transcribed in batch, with speaker detection.
Dutch–English meetings with Whisper
Section titled “Dutch–English meetings with Whisper”Download Whisper Large V3 (Multilingual) in Settings → Transcription and select it. The download is about 3.1 GB. It runs on your Mac’s GPU and can show the transcript during recording. After-recording transcription and audio imports also use Loofah’s speaker detection. No transcription API key is required. Whisper is available in the desktop app; CLI transcription currently supports Soniqo models.
For Dutch meetings with English technical terms, set Dutch as your main language and add English under Additional spoken languages. Whisper transcribes speech in its original language, with translation disabled. Recognition of mixed-language speech still depends on the recording and model.
Set the main language
Section titled “Set the main language”Open Settings → App → Language & Region and choose Main language.
This language is used for generated summaries and titles. Changing it does not translate an existing transcript. Regenerate a summary if you want an older meeting summarized in the new language.
Add spoken languages
Section titled “Add spoken languages”Use Additional spoken languages when people may switch languages during a meeting. The app passes the full selection to the transcription model.
Only add languages that are actually likely to be spoken. Every selected language must be supported by the model, and adding a non-English language changes the streaming model from live to after-recording transcription.
Use Customize meeting summaries to control summary structure and wording.