openai / openai/codex

Live voice transcription disappears while speaking after a recent update (user report)

Open
#42,729 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

app bug
Dominant language
Rust
Stars
125k
Forks
19.4k
PR merge metrics
PR metrics pending

Description

What version of the Codex App are you using?

The connected Windows host reports installed Codex package 26.901.4073.0 on 2026-09-04. The user noticed this after a recent update; the exact last-good version and the precise client/surface displaying the voice transcript have not yet been established.

What subscription do you have?

Not included in the public report.

What platform is your computer?

Codex host: Windows x64. Whether the affected voice UI is local or remote is not confirmed; please do not assume the host identifies the voice client's platform.

What issue are you seeing?

The user reports that the live transcription text disappears partway through speaking after a recent update. They explicitly clarified that the affected text is the in-progress transcription while they are still speaking, not a report of already-sent conversation history being deleted.

This is a user-observed symptom, not yet independently reproduced. We have not established whether the audio or final transcript is lost, whether the text later returns, or whether this is a rendering, transcription-stream, or connection problem. No specific UI action or network interruption was reported as the trigger.

What steps can reproduce the bug?

Reported sequence (frequency and exact trigger still unknown):

  1. Begin speaking using the voice/transcription interface in a Codex conversation.
  2. Observe recognized text being displayed while speaking.
  3. The displayed transcription disappears partway through continued speech.
What is the expected behavior?

Already-recognized in-progress text should remain visible and recoverable, or the UI should explicitly explain any replacement, reconnection, or transcription failure. Silent disappearance leaves the user unable to verify what has been captured.

Please investigate the recent-update regression, explain a supported recovery/workaround, and clarify whether content is retained. Do not require users to repeat long dictation simply to recover from a UI failure.

Additional information

Related but not confirmed identical:

  • #30862: dictation stops and returns to an empty input on Android.
  • #36537: drafts/transcription disappear during Android Remote connection interruptions.
  • #28765: pending dictation lost after ASR failure and window/thread switching on macOS.

The new report concerns text already visible during ongoing speech; the client platform and failure trigger need confirmation. No private audio, dictated content, program identity/purpose, local paths, or session identifiers are included. This symptom is not asserted to share a cause with #42728.

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

No files, tests, or entry points are identified. First confirm the affected client surface and reproduce the disappearance after beginning speech, then determine whether visible in-progress text is retained, replaced, or lost. Done means the regression cause, supported recovery or workaround, and content-retention behavior are established.

Written by the indexing model from the issue text.

Assessment

Domain
frontend
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Active
Clarity
Needs clarification
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.