Voice mode is a powerful feature, but it currently has two frustrating limitations that hurt the overall experience:
- Interruptions During Speech: Right now, voice mode often cuts off users mid-sentence. It would be a huge improvement to let users choose a specific end-word (like “over” or “done”) to signal the end of their message. This would allow for more natural, uninterrupted communication, especially for longer thoughts or when speaking more slowly.
- Lost Transcripts on Connection Issues: When there’s a connection hiccup, voice mode often responds with “Sorry, I didn’t catch that” and erases everything the user just said—even if it was several minutes long. It’s frustrating and discourages longer interactions. Ideally, the system should:
- Temporarily cache audio locally until transmission is confirmed.
- If connection is lost, attempt to transcribe locally and paste the result into the chatbox for manual sending.
- At minimum, give the user the option to retrieve what was just said instead of discarding it entirely.
Both of these improvements seem technically feasible and would dramatically enhance usability and trust in voice mode. Thank you for considering this request!