deniz.in

Markets

Weather

Loading weather

· via The Verge

ChatGPT gains Live Dictation with real-time transcription and no recording cap

OpenAI is rolling out Live Dictation for ChatGPT, which transcribes spoken prompts in real time across all plans, alongside the removal of the 10-minute dictation limit.

ChatGPT gains Live Dictation with real-time transcription and no recording cap

Live Dictation transcribes while you speak

OpenAI has started rolling out Live Dictation for ChatGPT, a feature that converts spoken prompts into text as the words are spoken, rather than making the user record everything first and transcribe afterward. According to The Verge, the rollout is global and covers every ChatGPT plan, from the free tier upward, across the web interface, the desktop client, and the iOS and Android mobile apps.

How it works

The upgrade applies to ChatGPT's dictation input, the option that lets people speak a prompt instead of typing it. Until now, audio had to be captured in full and processed before any transcript appeared. With the new behavior, transcription happens live while the speaker is still talking.

The Verge reports that during dictation, tapping or clicking the waveform display reveals the transcript as it forms. When the speaker finishes, the text sits in the prompt box, where it can be edited before submission. That editing step matters for anyone who has watched a transcription engine mangle a name, an acronym, or a technical term: errors become visible immediately and can be corrected before the model ever receives the prompt.

No more ten-minute cap

Two further changes ship alongside the feature. Per The Verge, OpenAI has dropped the 10-minute limit that previously applied to voice dictation sessions, removing a constraint for anyone using speech for longer inputs such as notes or draft documents. The company also says dictation has become more dependable, though it has not published specifics about what was improved.

Part of a broader audio push

Live Dictation is not an isolated addition. The Verge notes that it follows a reworking of ChatGPT's two-way conversational voice mode earlier this year — the separate experience in which users talk with the assistant and hear spoken responses back.

There is also record mode, announced last year, which transcribes voice notes and meetings. That capability remains in beta and is limited to Plus, Pro, Enterprise, Edu, and Business workspaces, according to The Verge. Users outside those tiers still have a workaround of sorts: dictating through the prompt box lets anyone ask ChatGPT to transcribe a voice message or condense one into bullet points, though that works as a per-request task rather than a dedicated mode.

Why it matters

Real-time feedback changes the practical value of voice input. When a transcript appears as you speak, you can confirm the system heard you correctly, catch mistakes instantly, and fix them before sending — something the old capture-then-transcribe flow made awkward. Combined with the removal of the recording cap, dictation in ChatGPT becomes viable for longer, more deliberate writing tasks, not just quick one-line queries.

The launch also signals that OpenAI is investing in the text-prompt path to voice input as distinct from its conversational voice mode. Dictation keeps users inside the standard chat workflow — editable prompts, copyable text, existing context — while voice mode offers a spoken, hands-free experience. Strengthening both suggests OpenAI treats speech as a primary interface rather than a convenience add-on. And for people who type slowly, have accessibility needs, or simply think better out loud, a faster and uncapped dictation pipeline lowers the friction of getting ideas into ChatGPT at all.

  • #openai
  • #chatgpt
  • #voice-transcription
  • #speech-recognition
  • #ai-assistants

Related posts