Hello ChatGPT team,
I’m a long-time mobile user and a big fan of the voice input feature. However, the current version has a serious usability problem: when there’s a network error during voice transcription, the entire recording is immediately discarded — without any way to retry or recover it.
This is very frustrating because sometimes I speak for a whole minute, and due to just one moment of unstable internet, everything I said gets wiped out. There’s no retry, no caching, nothing.
By contrast, the older versions allowed the voice message to be temporarily saved, and once the network resumed, it could be transcribed again. This was far more forgiving and practical, especially for users with inconsistent connections.
I’m suggesting a small but powerful change:
Please add a manual retry button when transcription fails, instead of automatically discarding the recording.
Or even better:
Keep the voice cached locally until transcription is confirmed successful, then clear it. This would save users from losing important thoughts and improve the overall experience.
Thank you for considering this feedback!
— Yong Qing Tan
ChatGPT Plus Subscriber