Lost functionality in new Speech-to-text update

ChatGPT Feature requests
1

I used to use the speech-to-text feature as a speech-driven alternative to typing. Now it functions like a quick voice query tool, which removes flexibility.

Previously, I could:

  • Preview and edit the transcribed text before sending.
  • Correct speech-to-text errors (e.g., “WOKE” being transcribed as “WORK”).
  • Break longer messages into manageable chunks, such as one paragraph at a time.
  • Mix typing and voice input within the same message.
  • Pause mid-thought, think for a moment, then continue dictating.
  • Scroll through the text I had already spoken before sending.
  • Avoid crashes by composing in smaller parts.

Now I’m forced to:

  • Send voice input immediately without reviewing or editing.
  • Choose between typing or speaking, since typing is lost when switching to voice.
  • Speak the entire message in one uninterrupted take, with no pause/resume ability.
  • Risk losing everything if the app crashes during long input.
  • Submit or cancel—no middle ground for review or changes.

Suggested improvements:

  • Add a preview/edit step before submitting voice input.
  • Allow pause and resume functionality.
  • Let users combine typed and spoken input.
  • Provide a confirm-before-send option for speech-to-text.

Thanks

2

I agree. I am also really disliking the update to the speech to text.

If I want to edit an earlier message and add more text via voice, it now replaces the earlier message instead of adding to it.

In my opinion, it should insert the text where my cursor is, not make a new message.

The previous functionality allowed for voice to text to be even more efficient than typing manually.

Note: im on the Android app