[Feature Request] Add Toggle to Enable Pre-Send Speech-to-Text Correction Previe

ChatGPT Feature requests
, , ,
1

Title: Add Toggle to Enable Pre-Send Speech-to-Text Correction Preview

Problem:

Current voice dictation on the ChatGPT iPad iOS app version immediately submits the prompt after speech input ends, with no opportunity to review or correct transcription errors. This change was recently made, I would assume, due to the lack of oversight by product team leaders, or just plain sloppiness of the UX Dev Team.

But the lack of this functionality now leads to unintentional phrasing, incomplete thoughts, or STT misfires being submitted instantly—especially frustrating for high-frequency, high-context users, as well as wasting precious time, GPU cycles and power energy by the removal of this feature.

Request:

  1. Make the OpenAI ChatGPT app user interface consistent on all platforms, and restore the lost ability to edit translation errors beignet being sent to assistant on the iPadiOS version immediately*

  2. Add a user-configurable toggle “Automatic Prompt Send (Yes/No) in user settings for all versions

When disabled, the dictated text appears in the input box and does not auto-send—allowing the user to manually correct his prompt, or the multiple Apple Dictation STT errors common among all users, as well as modify prompt instructions before submission to assistant.

Why it matters:

  • Avoids unnecessary AI cycles spent deciphering bad transcription.

  • Dramatically improves prompt clarity and user control.

  • Especially useful for power users who dictate fast, long, or layered prompts.

  • A simple toggle satisfies both user types: those who want speed vs. those who want precision.

UX Reference:

This behavior mirrors the dictation preview experience on iOS Notes or Gmail, where users see the text before confirming. The lack of it in ChatGPT voice mode is a step backwards for dictation fidelity.

1 Like