OpenAI rolls out Live Dictation for ChatGPT, transcribing speech in real time
OpenAI has launched Live Dictation, letting ChatGPT transcribe spoken prompts as you speak and removing the 10‑minute limit.
Source: The Verge — AIOpenAI has introduced Live Dictation, a feature that transcribes spoken input in real time instead of waiting for you to finish recording. The change removes the previous ‑minute cap and is being deployed worldwide across all ChatGPT plans, which matters for anyone who relies on voice interaction to draft prompts, take notes, or generate content.
What is Live Dictation and how does it work?
According to The Verge, Live Dictation captures audio from your microphone and streams a live text transcript as you speak. While dictating, you can click or tap the audio waveform to view the evolving transcription and, once you finish, edit the text directly in the prompt box before sending it to the model. The system updates the transcript continuously, so you see words appear almost instantly.
Which platforms and plans receive the feature?
The rollout is global and applies to the web interface, the desktop application, and the iOS and Android apps. OpenAI says the feature is available on all ChatGPT plans, meaning free users, Plus subscribers, and enterprise customers all get the same live‑dictation experience. Earlier this year OpenAI updated the two‑way conversational voice mode, and Live Dictation builds on that foundation.
How does removing the ‑minute limit change usage?
Previously, voice dictation stopped after minutes, forcing users to break longer recordings into chunks. The new limit removal lets you dictate for as long as needed without interruption, which is especially useful for transcribing meetings, lectures, or brainstorming sessions. OpenAI also notes that the transcription engine is now "more reliable," suggesting fewer drop‑outs and higher accuracy over extended periods.
What are the practical implications for creators, developers and researchers?
- Content creators can dictate scripts, outlines, or social‑media copy on the fly, reviewing the live transcript for errors before the model processes it. This speeds up the iterative workflow of drafting and refining text.
- Developers building voice‑enabled applications can prototype interactions directly in ChatGPT, using the live transcript to fine‑tune prompts that later become part of their own products.
- Researchers can capture spoken notes or interview excerpts without needing a separate transcription service. The ability to edit the transcript on the spot means you can correct mis‑recognitions before they propagate into downstream analysis.
OpenAI also mentions a "record mode" introduced last year for transcribing voice notes and meetings. That mode remains in beta and is limited to Plus, Enterprise, Edu, Business, and Pro workspaces. In addition, the standard prompt‑box dictation can be used to ask ChatGPT to turn voice messages into bullet‑point summaries, providing a quick way to distill spoken information.
How can you experiment with Live Dictation concepts in Graf Grafy?
Grafy does not yet host OpenAI’s Live Dictation natively, but you can replicate a similar workflow using existing apps:
- Capture audio with any recorder on your device.
- Upload the file to Grafy Files (/files), which supports audio‑to‑text conversion via integrated cloud models. The transcription can be edited directly in the file viewer.
- Copy the resulting text into Grafy Chat (/chat) to interact with a variety of LLMs side‑by‑side. Because Chat lets you switch models, you can compare how different systems respond to the same spoken prompt.
- (Optional) Track prompt iterations in Grafy Graph (/graph) to visualize how changes to the transcript affect model outputs over time.
While this approach requires a manual upload step, it lets creators, developers, and researchers experiment with real‑time dictation ideas without leaving the Grafy workspace. You can also use the Files app to batch‑process multiple recordings, then feed each transcript into Chat for rapid prototyping.
Source: The Verge – "ChatGPT can transcribe as you speak now with Live Dictation" (Nov 19 2026).
Frequently asked questions
Does Live Dictation work on the free ChatGPT plan?
Yes. OpenAI states the feature is rolling out globally on all ChatGPT plans, which includes the free tier.
Can I still use the older "record mode" for meetings?
The record mode remains in beta and is limited to Plus, Enterprise, Edu, Business, and Pro workspaces. It is separate from Live Dictation, which is available to everyone.
How do I edit the live transcript before sending it to the model?
While dictating, click or tap the audio waveform to reveal the live text. Once the transcription appears, you can modify the text directly in the prompt box and then submit the edited version.
Source: The Verge — AI — ChatGPT can transcribe as you speak now with Live Dictation


