Guides·Published

Can ChatGPT Transcribe Audio? Limits and What to Do Instead

Yes, but only on a paid plan and only for short recordings. OpenAI's help page says uploaded recordings can be up to 6 minutes long on GPT-6 Instant, audio files can be up to 512 MB, and video files are not accepted.

We read OpenAI's help page on 15 November 2026 (the page showed it was updated the day before). Everything below about ChatGPT comes from that page: Uploading files and audio to ChatGPT. PlainScribe is our product, and it is one of the options for longer files.

TL;DR

  • Paid plans only. "Audio uploads are not available on the Free plan at this time," says OpenAI.
  • 6 minutes per recording on GPT-6 Instant in chat. Longer recordings may be split into sections when Data Analysis is available, but that is best-effort and very long recordings may time out.
  • 512 MB per audio file.
  • No video. WebM and MP4 files that ChatGPT identifies as video are rejected. Audio-only MP4 and WebM work.
  • For a clip under 6 minutes on a paid plan, ChatGPT is a fine choice. For a meeting, lecture or interview, use a tool built for long files.

ChatGPT's audio limits

Checked 15 November 2026 on OpenAI's help page.

LimitWhat OpenAI's page says
PlansPaid ChatGPT subscriptions and workspaces. Not the Free plan.
Length per recordingUp to 6 minutes on GPT-6 Instant in chat
Longer recordingsMay be processed in smaller sections when Data Analysis is available; best-effort, may time out
File sizeUp to 512 MB per audio file
Audio formatsWAV, MP3/MPEG, OGG/OGA, audio-only WebM, PCM, FLAC, AAC, M4A, audio-only MP4
VideoNot supported: WebM and MP4 files identified as video are rejected
Where it worksNot in Work or Codex
Accuracy"Transcripts may contain errors, and speaker identification may be unreliable."
Upload volumeGeneral file uploads: up to 80 files every 3 hours

The 6-minute limit is the one that matters. Most recordings people want transcribed are meetings, lectures, interviews and podcasts, and those run 30 minutes to several hours.

When ChatGPT is the right tool

  • You pay for ChatGPT and the clip is under 6 minutes: a voice memo, a short voicemail, a quote from an interview.
  • You want to do something with the words straight away, such as summarize them, translate them or turn them into an email, in the same chat.
  • The file is audio, not video, or you can export just the audio.

In those cases there is no reason to sign up for anything else.

When to use something else

  • You're on the Free plan. Audio uploads aren't available. The free Gemini app accepts up to 10 minutes of audio in total (Google help).
  • The recording is longer than 6 minutes. ChatGPT may split it into sections, but OpenAI calls that best-effort. A 1-hour recording is 10 times the stated limit.
  • It's a video file. Export the audio track first, or use a tool that accepts video.
  • You need reliable speaker labels or timestamps. OpenAI warns that speaker identification may be unreliable.

What to do with a longer recording

Pick a tool whose per-file limit is longer than your file. Limits checked 15 November 2026 on each vendor's page:

Your recordingOptions
Up to 10 minutesGemini app, free (Google)
Up to 30 minutesTurboScribe Free, 3 files a day (TurboScribe); Otter Basic, 30 minutes per conversation and 3 imports in total (Otter)
Up to 3 hoursGemini app on Google AI Pro or Ultra (Google)
Up to 10 hoursTurboScribe Unlimited, $120 a year (TurboScribe); Notta Pro, 5 hours per recording, $97.99 a year (Notta)
Any length, paid onceSonix Pay As You Go at $10 an hour (Sonix); PlainScribe at $4 an hour, with no length cap and uploads up to 1 GB

If you only transcribe once in a while, paying per hour avoids a subscription. If you transcribe hours every month, a flat plan such as TurboScribe Unlimited costs less. Our guide to transcribing a long audio file has the cost of a 2-hour file on each tool, and transcription upload limits compared lists every plan's caps.

Then bring the transcript back to ChatGPT

A transcript is text, and ChatGPT handles long text well. A common workflow:

  1. Transcribe the full recording in a tool built for long files.
  2. Export it as TXT or Markdown.
  3. Upload the text to ChatGPT and ask for a summary, action items, quotes or a translation.

This gets you past the 6-minute audio limit and keeps ChatGPT for what it does best.

What about the OpenAI API?

Developers can send audio to OpenAI's speech-to-text API instead of the ChatGPT app. Its own limit is lower: "Files can be up to 25 MB," per OpenAI's speech-to-text guide. Longer files have to be split or compressed first.

FAQs

Can ChatGPT transcribe audio for free? No. OpenAI's help page says audio uploads are not available on the Free plan. The free Gemini app accepts up to 10 minutes of audio.

How long can an audio file be for ChatGPT? Up to 6 minutes on GPT-6 Instant in chat, and up to 512 MB. Longer recordings may be processed in sections when Data Analysis is available, but OpenAI says that is best-effort and may time out.

Can ChatGPT transcribe a video? No. WebM and MP4 files that ChatGPT identifies as video are not supported for audio uploads. Export the audio track as MP3 or M4A first, or use our free video to audio extractor.

Can ChatGPT transcribe a 1-hour meeting? Not reliably. One hour is ten times the 6-minute limit. Transcribe it with a tool that accepts long files, then paste the transcript into ChatGPT for the summary.

Does ChatGPT label speakers? OpenAI's page says speaker identification may be unreliable. If you need to know who said what, use a transcription tool with speaker labels.

Summary

ChatGPT transcribes short audio clips on paid plans: up to 6 minutes per recording, 512 MB, audio only. For anything longer, transcribe the file in a tool built for long recordings and bring the text back to ChatGPT. See how to transcribe a long audio file for the options and costs.

Transcribe, Translate & Summarize your files

Your first transcript is free (up to 1 hour). No credit card required.