AI

ChatGPT Live Dictation: text appears as you speak, no time cap

Key takeaways

  • Live Dictation shows your words as you speak instead of after you stop
  • The 10-minute recording limit has been removed
  • It is rolling out globally to all ChatGPT plans on iOS, Android, web and the desktop app
  • Tapping the waveform shows or hides the live text

What is ChatGPT's Live Dictation, and what changes for people who talk to it rather than type? OpenAI has updated the speech-to-text feature in ChatGPT so that words appear on screen as they are spoken, and it has removed the 10-minute recording limit that used to cut long dictation short.

What Live Dictation actually is

Dictation in ChatGPT converts your voice into a typed message. Before this update, users spent more time waiting for their message to finish transcribing. Live Dictation changes that to a streaming model, where the text builds up while you are still talking.

OpenAI described it on X as making dictation "faster and more reliable". In the company's words, "your speech turns into text as you talk, so you spend less time waiting for your message to finish transcribing." Atty Eleti, who works on voice features at OpenAI, confirmed that the 10-minute dictation limit has also gone as part of the streaming upgrade.

How it works

Streaming dictation sends the audio in a continuous flow, and the transcript comes back in pieces, so you see each phrase appear shortly after you say it.

The release notes describe one control: tap or click on the waveform to show or hide the text as you go. That gives users a choice. Someone who wants to check wording as they speak can watch the transcript form. Someone who prefers to ramble through a thought without being distracted can hide it and read the result afterwards.

Who uses it and why

Anyone drafting a long prompt benefits most. A person briefing ChatGPT on a complicated problem, such as describing a project, a bug or a set of meeting notes, could previously hit the 10-minute ceiling mid-thought. With the cap removed, a long monologue can now be captured in one go. The faster feedback also helps when the transcript is wrong: spotting a misheard word as it appears is quicker than discovering it at the end.

What the limitations are

The reporting is thin on detail, and several practical questions remain open. Neither source says how accurate streaming transcription is compared with the previous approach, how it handles accents or noisy rooms, or whether there is any new cap on length in its place. The 10-minute limit is described as removed, but the sources do not say whether any new cap applies.

Reliability is the claim OpenAI makes, but it is a claim from the company's own announcement and release notes, not an independent test. Users who rely on dictation for anything important should check the output before sending it, as with any speech-to-text tool.

What this means in practice

Live Dictation is rolling out globally to all ChatGPT plans on iOS, Android, web and the desktop app. The source gives no UK-specific pricing or availability notes. For UK and US professionals who use ChatGPT for drafting, the change is a small but real time saver: no waiting after you finish speaking, and no need to split a long dictation into chunks.

Updating the app is the first thing to try. The waveform tap is the control to look for.

Where this is heading

Streaming transcription is a basic building block for that: once text appears in real time, the gap between speaking and typing narrows. The next things to watch are independent accuracy comparisons and whether OpenAI publishes any new limits or usage rules for long recordings.

More from Future Technology HQ