Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.

### Answer Based on the available documentation, the Custom LLM WebSocket implementation does not explicitly provide a language field in the transcript or conversation data sent to your server [(1) ]() . The events you receive from Retell include the

transcript

field containing utterances with

role

and

content

, but there's no documented language identifier in these events . The transcript objects contain the spoken content but don't specify which language was detected . For multilingual support in Custom LLM implementations, you would need to implement language detection on your server side by analyzing the transcript content yourself, or handle it at the prompt level by instructing your LLM to detect and respond in the appropriate language [(2) ]() .

​ To learn more, see the sources I considered: [[1] LLM WebSocket](https://docs.retellai.com/api-references/llm-websocket) [[2] Discord message from @jeffersoncooper_35612 on 2025-06-18](https://discord.com/channels/1186424376300679209/1186462578310910112/1384978764295045337) ​