Bot Language
The lang parameter when creating a new bot is pretty vague, we don't provide a certain format to work in or even specific options. This is because this lang parameter is solely used when transcribing your recording. Therefor the lang parameter must conform the requirements of the selected transcription model (if the model supports language selected). The lang parameter is optional, if not present we'll always resort to the "language detection system" of the transcription model.
When using realtime models
The lang parameter is required when using realtime transcription models. Not providing a lang parameter when starting the bot with a realtime model will lead into the realtime websocket stream not working properly.
Whisper
- Requires full English language names (e.g., "English", "Spanish", "French")
- Full list available in the OpenAI Speech to Text Documentation
AssemblyAI
- Supports 102 languages using language codes (e.g., "en", "es", "fr")
- Full list available in the AssemblyAI Documentation
Deepgram
- Supports 37 languages using language codes (e.g., "en", "es", "fr")
- Full list available in the Deepgram Documentation
Speechmatics
- Supports many languages using language codes (e.g., "en", "es", "fr")
- Full list available in the Speechmatics Documentation
Rev AI
- Supports many languages using language codes (e.g., "en", "es", "fr")
- Full list available in the Rev AI Documentation
- Our deployment region is EU, so only EU-available languages are supported
ElevenLabs
- Supports 99 languages using 3-letter codes (e.g., "eng", "spa", "fra")
- Full list available in the ElevenLabs Documentation
Gladia
- Supports many languages using 2-letter codes (e.g., "en", "es", "fr")
- Full list available in the Gladia Documentation