OpenAI has announced an upgraded “Live Dictation” feature for ChatGPT, significantly improving speech-to-text speed and stability and formally removing the previous 10-minute recording limit. The update has been rolled out globally to all free and paid plan users across iOS, Android, web, and desktop apps.

ChatGPT upgrades its “real-time dictation” feature, fully supporting transcription while speaking and unlimited duration.
Previously, when users dictated long prompts or notes in ChatGPT, they were not only limited by the 10-minute recording cap but also had to wait until the entire recording was processed before seeing the text result. The upgraded real-time dictation completely changes this workflow—now, as soon as you start speaking, the text appears on the screen in sync. This “what you say is what you get” transcription experience lets users catch recognition errors in real time and adjust wording at any moment. Whether brainstorming for long periods, writing a first draft, or dictating complex multi-step instructions, they can do it all in one go without interruption.

Although most operating systems include basic voice input tools, ChatGPT’s real-time dictation is deeply integrated directly into the chat interface, seamlessly connecting the interactive experience of “speaking aloud, reviewing text, and sending commands.” For creators, developers, and general users who are used to sorting through complex thoughts orally, this feature transforms one-way voice recording into a smooth, continuous real-time editing experience.

In terms of interface design, OpenAI also took into account that some users can easily be distracted by moving text while brainstorming, and added a thoughtful interactive control: during recording, users can simply click or tap the “audio waveform” on the screen to hide or show the live transcription text at any time.
Dictation in ChatGPT is getting faster and more reliable. With Live Dictation, your speech turns into text as you talk, so you spend less time waiting for your message to finish transcribing. Yap away! https://t.co/sav9p2fyyu
— ChatGPT (@ChatGPT) November 18, 2026
Although OpenAI did not disclose specific latency benchmark data, accuracy metrics, or a technical maximum duration limit in this update, and only emphasized that the new system is “faster and more reliable” than its predecessor, it gave privacy-conscious users clear control. Users can go to “Data Controls” in ChatGPT settings at any time to turn off “Include audio recordings” or disable the “Improve the model for everyone” option, freely managing the sharing permissions for voice data.
The upgraded real-time dictation is now gradually rolling out across platforms worldwide. Users can simply click the microphone icon next to the input box to immediately experience efficient hands-free input with no time limits.
Source: KOCPC Chinese