Hey, thanks for the report. This is a known design issue with voice mode’s auto-translation behavior.
You’re not alone, we’ve seen the same request for Chinese: Supports Chinese output in the voice input mode
Your suggestion to keep the original language visible before translation is excellent. I’ll pass this to the team and link your report and the Chinese user’s report so engineering can track it.
For now, you’ll need to type manually to ensure accuracy when translation matters while communicating with the AI.
Voice mode was just released with Cursor 2.0, so this kind of feedback is very helpful for improving it.