Feature request for product/service
Cursor IDE
Describe the request
When I dictate with the microphone in the Agent/Composer input, the live streaming transcript appears in the text field while I speak.
This is distracting:
I focus on reading/correcting the live text instead of thinking about what I want to say.
Because the interim text jumps/rewrites, I sometimes feel that Cursor didn’t capture what I said, so I stop mid-sentence, repeat myself, or cut myself off.
Previously, dictation felt calmer: waveform / listening UI only, and the full transcript appeared after I stopped. That matched how I work better.
Accuracy after finalize can be fine — the issue is the live interim display interrupting concentration.
Expected / requested behavior
Please add a setting, for example:
Hide live transcript while dictating / Show interim transcript: off
Or restore optional batch / finalize-on-stop mode: waveform only while speaking, insert text when recording stops.
Default can stay as streaming for users who like it; I just need an opt-out.
Why it matters
Voice input should reduce friction. Live interim text increases cognitive load and makes me second-guess the STT mid-utterance, which leads to worse prompts and more restarts.
Operating System (if it applies)
MacOS