New video: 8 Real Jobs AI Agents Can Do In Your Browser →
Muddaser

ChatGPT Live Dictation: Setup, No 10-Minute Cap & Fixes

Quick answer: ChatGPT Live Dictation turns speech into editable text while you are still talking. OpenAI says the update removes the previous 10-minute recording cap and is rolling out globally to every ChatGPT plan on iOS, Android, web, and the desktop app.[4] Tap the microphone in the message box, speak, review the text, correct names or numbers, and then send it.

What changed in ChatGPT Live Dictation?

OpenAI announced the update on November 18, 2026. Three changes matter:

  • Text appears while you speak. You no longer have to wait until the recording ends to see the full transcription.
  • The running transcript can be shown or hidden. Tap or click the waveform while recording.
  • The old 10-minute cap is gone. OpenAI has not announced a replacement maximum, so “no 10-minute cap” is more accurate than calling dictation unlimited.[4]

The rollout covers Free and paid plans on iOS, Android, the web interface, and the ChatGPT desktop app. Because OpenAI describes it as a rollout, the updated interface may not reach every account at the same moment.[4]

How to use ChatGPT Live Dictation

  1. Open ChatGPT on the web, desktop, iPhone, iPad, or Android.
  2. Start a new chat or open the conversation where you want to add a message.
  3. Select the microphone icon in the message box.
  4. Allow microphone access if your browser or device asks.
  5. Speak naturally. Select the waveform if you want to show or hide the live text.
  6. Stop the recording when you finish.
  7. Review the returned text, especially names, dates, prices, URLs, and technical terms.
  8. Edit anything that is wrong, then send the message.

OpenAI’s dictation documentation confirms that the audio is sent to its models for transcription and that the returned text can be edited before it becomes your message.[6]

Live Dictation vs ChatGPT Voice

These controls solve different problems:

  • Use Dictation when you want to speak one prompt, inspect the exact text, edit it, and send it.
  • Use Voice when you want a live back-and-forth conversation in which ChatGPT also speaks.

OpenAI describes Voice as a spoken conversation and Dictation as a single recording converted into editable text. It also warns that Voice transcripts are not necessarily verbatim, which is another reason to prefer Dictation when exact wording matters.[5]

A 60-second accuracy test

Before dictating an important email, report, or instruction, run this short test:

  1. Say your full name and two uncommon names you use at work.
  2. Dictate one date, one decimal, one currency amount, and one web address.
  3. Include a correction while speaking: “Tuesday—correction—Wednesday.”
  4. Stop and compare the transcript with what you intended.
  5. Fix recurring errors by spelling a name or breaking a long number into smaller groups.

Do not send dictated legal, financial, medical, or client-facing text without reviewing it. Speech-to-text can still mishear a proper noun or turn a small pause into a different sentence.

Five useful dictation templates

You can dictate the raw information first and ask ChatGPT to structure it afterward.

1. Client follow-up

Turn these spoken notes into a concise client follow-up. Separate decisions, actions for me, actions for the client, and deadlines. Do not invent missing details; list questions at the end.

2. Meeting notes

Organize this brain dump into summary, decisions, owners, due dates, risks, and unresolved questions. Preserve every date and number exactly as dictated.

3. Long email

Rewrite this as a clear email with a direct subject line, a two-sentence opening, short paragraphs, and a specific next step. Keep the facts unchanged.

4. Article outline

Convert these ideas into an outline. Put the direct answer first, group duplicates, flag unsupported claims, and create an FAQ from the real questions I mention.

5. Technical issue report

Turn this dictation into a reproducible bug report with environment, expected result, actual result, exact steps, errors, and troubleshooting already attempted. Do not guess any missing version numbers.

ChatGPT Dictation not working: checks in order

1. Confirm that you selected Dictation

Use the microphone control that records a message and returns editable text. If ChatGPT starts speaking back immediately, you likely opened Voice instead.

2. Check the rollout and update the client

Update the ChatGPT mobile or desktop app, or reload ChatGPT in a current browser. The feature is rolling out, so an account that still shows the older dictation experience may simply not have received the new interface yet.[4]

3. Re-enable microphone permission

On the web, use the site-permission control beside the address bar and allow microphone access for ChatGPT. In a mobile or desktop app, check the operating system’s privacy settings and confirm that ChatGPT is allowed to use the microphone.

4. Test the microphone outside ChatGPT

Record a short clip in your device’s built-in recorder. If the input meter does not move or the recording is silent, fix the selected microphone, hardware mute, Bluetooth route, or system permission before troubleshooting ChatGPT.

5. Remove browser-specific variables

If the microphone works elsewhere, try a private window or another current browser. Grant microphone permission in that test window. If dictation works there, inspect extensions and the normal browser profile rather than resetting the whole device.

6. Check service status

If dictation fails across multiple devices or networks at the same time, check OpenAI’s status page before reinstalling apps or clearing data.

Privacy and audio retention

OpenAI says dictation audio is retained while the related chat remains in your chat history. When you delete the chat, OpenAI says it deletes the associated audio clip within 30 days, with stated exceptions for security or legal obligations and for audio previously shared for model training after being disassociated from the account.[6]

Consumer users can stop sharing new dictation audio for model improvement under Settings → Data Controls by disabling Include your audio recordings, or by turning off Improve the model for everyone. OpenAI says business-user inputs and outputs are not used for training by default.[6]

For sensitive material, the safer workflow is simple: do not dictate secrets, passwords, payment-card details, private medical information, or confidential client data unless your organization has approved that use.

Frequently asked questions

Is ChatGPT Live Dictation available on the Free plan?

OpenAI says it is rolling out globally to all ChatGPT plans, including on iOS, Android, web, and desktop.[4]

Did OpenAI remove the dictation time limit?

OpenAI removed the previous 10-minute recording limit. It has not stated that there can never be another technical or account limit, so describe the change as “no 10-minute cap,” not “unlimited.”[4]

Can I edit the transcript before sending it?

Yes. OpenAI says the transcription is returned as editable text before it is sent as your message.[6]

Is Live Dictation the same as Voice mode?

No. Dictation produces an editable text prompt. Voice is intended for a live spoken conversation with ChatGPT.[5]

What should I check in every transcript?

Review names, dates, currency amounts, decimal points, email addresses, URLs, product names, and any sentence where a pause could change the meaning.

Sources

[4] https://help.openai.com/en/articles/6825453-chatgpt-release-notes — ChatGPT release notes
[5] https://help.openai.com/en/articles/8400625-voice-mode-faq — Voice Mode FAQ
[6] https://help.openai.com/articles/12168547 — ChatGPT Dictation