Multilingual Dictation: 25+ EU Languages in One Model
English, German, French, Spanish, Italian, Portuguese, Dutch, Polish — switch languages mid-sentence without changing settings.
SuperSpeech uses a single multilingual model (NVIDIA Parakeet-TDT-0.6B-v3) that handles 25+ European languages natively. You do not need to select a language, install a language pack, or toggle settings when switching from a German email to an English PR description. The model detects the spoken language automatically and transcribes accurately.
Supported languages
Full support for these European languages in a single download: English, German, French, Spanish, Italian, Portuguese, Dutch, Polish, Czech, Slovak, Romanian, Hungarian, Bulgarian, Croatian, Slovenian, Danish, Swedish, Norwegian, Finnish, Estonian, Latvian, Lithuanian, Greek, Maltese, Irish.
- No language packs to install — all 25+ languages ship in one model
- Automatic language detection at the sentence level
- Mixed-language input works (Denglisch, Spanglish, multilingual meetings)
- Custom dictionary entries per language
- Export transcripts with detected language metadata
Why multilingual matters in Europe
European professionals routinely work across languages. A German lawyer may dictate briefs in German, correspond with clients in English, and review contracts in French — all in the same session. SuperSpeech handles that switch transparently because the underlying model was trained on all supported languages simultaneously.
Accuracy benchmarks
Word Error Rate (WER) on clean audio is typically 5-8% in English, German, French, Spanish, and Italian. Less-resourced languages (Maltese, Irish) have slightly higher WER but remain usable. For challenging domains, custom dictionary entries can reduce WER significantly on domain-specific vocabulary.
Use cases
Multilingual dictation shines in specific workflows:
- Translators working in 2-3 language pairs daily
- Multilingual sales and customer support teams
- Academics writing in English but interviewing in local languages
- Healthcare serving patient populations with multiple languages
- International legal practice
Frequently Asked Questions
Do I need separate licenses for different languages?
No. Every SuperSpeech license includes all 25+ supported languages. One purchase covers every language you work in.
Can I mix languages in one session?
Yes. The model auto-detects language at the sentence level. You can dictate German, then switch to English, then switch to French — no setting changes.
Which European languages have the highest accuracy?
English, German, French, Spanish, Italian, Portuguese, and Dutch have the strongest training data. Polish, Czech, and Scandinavian languages are also strong. Less-resourced languages work but with slightly higher WER.
Does it handle accents and dialects?
Yes. The model is trained on diverse speaker populations. For specific regional accents, custom dictionary entries can help with commonly misrecognized terms.
What about non-European languages?
The current model focuses on European languages. Chinese, Japanese, Korean, and Arabic are on the roadmap but not yet supported.
Dictate in every language you work in
One license, 25+ languages, lifetime updates on the Lifetime plan.