ScribVoiceAudio transcription guide
Can ChatGPT Transcribe Audio?
Updated 5 November 2026
If you have a recording and want ChatGPT to turn it into text, the answer depends less on what ChatGPT can do than on the file you already have — and on whether your plan can upload audio at all.
The short answer
ChatGPT can transcribe audio when the upload is available to your plan and matches the documented formats. Three limits matter: audio uploads are paid-plan only, the supported list is specific, and long recordings are handled on a best-effort basis.
What ChatGPT actually supports
The OpenAI upload documentation lists these audio inputs: WAV, MP3/MPEG, OGG/OGA, audio-only WebM, PCM, FLAC, AAC, M4A, and audio-only MP4. The same documentation says audio uploads are available on paid plans, at 512 MB per file.
That list is narrower than the set of files people bring to an upload form. It also distinguishes audio-only WebM and audio-only MP4 from files that are identified as video. A video WebM or video MP4 is not accepted as an audio upload under that documented rule.
Where it stops
- Plan access: Audio uploads are available on paid ChatGPT plans.
- File and container limits: Each audio file is capped at 512 MB. Files outside the documented audio list need a different route before transcription can start.
- Long recordings and labels: Long recordings are processed in segments on a best-effort basis and may time out. The same documentation warns that transcription can contain mistakes and that speaker identification is not reliable.
The documentation does not list SRT or VTT subtitle export. That matters when the desired result is a timed caption file rather than only a block of recognized text.
Why a dedicated tool handles these
A dedicated upload workflow accepts the file types on its supported uploads list. ScribVoice's list includes .amr .wma .mkv .m2ts, so these can be uploaded directly — no MP3 copy needed. Because browsers cannot inspect these containers, each is capped at 200 MB on every plan, including Standard and Pro, and the no-account trial's 100 MB limit applies as well.
It can also accept video files, including MP4, MKV and M2TS, and transcribes the audio track — video preview is not available in the workspace. See MP4 to text.
The output is designed for the next step: ScribVoice supports SRT, VTT, DOCX, TXT exports. Speaker labels are included in the transcript when the language and recording support them, and the transcript text can be edited before export.
The workflow most people end up using
- Check whether the recording is already in ChatGPT's documented audio list and whether your plan permits audio uploads.
- If it is a WMA, AMR, MKV, M2TS, or another file outside that list, upload the original to a dedicated transcription workflow instead of making an extra conversion copy.
- Choose the spoken language when needed, let the transcript finish, review names and speaker labels against the recording, then export the format required by the next tool or person.
For other supported recording sources, see MP3 transcription, WAV transcription, or M4A transcription.
Bottom line
ChatGPT can be useful for supported audio uploads, but its documented plan, format, size, and long-recording limits decide whether it fits a particular file. If you already have a WMA, AMR, video MP4, or another supported ScribVoice format, you can start with the original recording and export a transcript or subtitle file.
Try 30 minutes free without an account on the ScribVoice transcription workspace.