Multilingual
SimplAI supports multilingual voice agents — calls that span English plus other languages, or non-English calls — by combining a multilingual ASR with a multilingual-capable TTS.
Enable multilingual ASR
Multilingual ASR is available on two providers in the Transcriber section:
- Deepgram → Multilingual model. In the Deepgram model selector, pick Multilingual. Deepgram routes the call through its multilingual transcription model.
- Google → Multilingual model. Google is available as a STT provider. Its model selector includes a multilingual option.
TTS providers are filtered automatically
When a multilingual ASR option is selected, the TTS provider dropdown filters to only show providers known to support multilingual text-to-speech input. Currently the qualifying providers are:
- ElevenLabs
- OpenAI
Providers that do not support multilingual TTS are hidden (not greyed out) when a multilingual ASR is active. If you switch back to a single-language ASR, the full list returns.
An inline note appears in the TTS section to explain the filter:
Multilingual input is active. Only TTS providers that support multilingual synthesis are shown.
Caveats
- Provider capabilities evolve. If a new provider adds multilingual TTS support and isn't yet in the filter list, raise it with your support contact to get it added.
- Other configuration carries over. Voice settings, fallback voices, BYOK keys, and voice cloning behave identically in multilingual mode — see Voice (TTS) and BYOK.
- Realtime API mode does not use this filter — the OpenAI Realtime API handles language natively. If you're on Realtime API mode, the multilingual choices in this page don't apply.