Can ChatGPT transcribe audio?
Yes, on paid plans, and with four limits worth knowing before you rely on it. ChatGPT will take an audio file and give you back the words. What it will not give you is speaker labels you can trust, timestamps you can click, a subtitle file, or a transcript tied to the audio so you can check a doubtful word. For a short voice note that is fine. For an interview, a lecture or a full episode it is not.
Upload a file, read the transcript, click any line to hear it.
What transcription looks like when the transcript stays linked to the audio
- 1
Check your plan and platform
Free ChatGPT accounts generally cannot upload an audio file for transcription. Paid plans can, and availability of Record Mode varies by platform. If the upload button is not there, this is why.
- 2
Ask for verbatim, not a summary
Left to itself ChatGPT will often summarise. Say the word verbatim. Record Mode in particular is built to produce a structured summary rather than a word-for-word transcript, which is the single most common surprise.
- 3
Split anything long
The Whisper API rejects a single request over 25MB, which a one-hour MP3 at 128 kbps will exceed on its own. Splitting works, but you then have to stitch the pieces back together and the speaker attribution will not carry across the seam.
- 4
Check what came back
Read it against the audio. Names, jargon and acronyms are where machine transcripts slip, and in ChatGPT you have no way to click a line and hear it, so every check means scrubbing through the file by hand.
Keep working inside the Hilite editor
Your uploaded files open directly in a focused editor view. Arrange your clips, preview the timeline, adjust transitions, and export when the track sounds perfectly right.
- Review the transcript in context
- Keep timestamps with the words
- Copy, export, or keep editing
Input formats
Output formats
Built for every kind of audio
Whatever you are making, this tool helps you get to a cleaner, more useful result, then hand off to Hilite to enhance and polish.
- It gives you prose, not a transcriptA transcript is text plus structure: who spoke, when they spoke, and a link back to the audio. ChatGPT returns the text and drops the structure, which is most of what makes a transcript useful.
- You cannot verify a wordEvery transcript has a handful of doubtful words. The question is how long it takes to check one. With a transcript linked to the audio it is one click. In a chat window it is a manual hunt through the file.
- It is a dead endOnce ChatGPT hands you the words, the recording and the text go their separate ways. Nothing you do to the transcript touches the audio.
Made for podcasters and their teams
Trim the dead air off the top and tail, and every recording is ready to edit, enhance, and publish.
FAQ
Common questions
If you do not see your question here, the support team replies within one business day.
- Generally no. Free accounts can use live dictation, which turns your speech into text as you talk, but uploading a recorded file for transcription is a paid-plan feature. This changes from time to time, so check the upload control in your own account.
- Not reliably. The base Whisper model has no speaker diarization at all, and where labels do appear they are often missing, swapped or too generic to use. For a two-person interview you will usually have to assign the speakers yourself.
- Not in a usable form. There is no SRT or VTT export, and any timestamps in the output are approximate rather than something you can click to jump to. If you need captions, you need a tool that produces subtitle files.
- Shorter than most people expect. The Whisper API limit is 25MB for a single request, which a one-hour MP3 at 128 kbps already exceeds. Anything longer has to be split into parts and reassembled by hand.
- Hilite transcribes in the browser with speaker labels, clickable timestamps and TXT, DOCX or SRT export, in 99+ languages, and every line stays linked to the audio so a doubtful word takes one click to check. It is free to start and there is nothing to install.
Need the parts ChatGPT leaves out
Speaker labels, clickable timestamps and subtitle export, free in the browser, with the transcript still tied to the recording.