How-to · Speaker diarization

Transcribe an interview with speaker labels

A wall of text is hard to use when three people spoke. The diarization route labels who said what and adds timestamps, so interview and meeting transcripts are readable and quotable.

Transcribe with speaker labels

In short

Choose GPT-4o Transcribe Diarize for speaker turns and timestamps.
Upload MP3, MP4, MPEG, MPGA, M4A, WAV or WebM up to 25 MB.
Add names and jargon as context to improve spelling.
The credit quote depends on model and length and is shown before you start.

Prepare the recording

One microphone per speaker gives the cleanest separation.
Avoid people talking over each other where possible.
Trim silence and off-topic sections to stay under 25 MB.

Add context

A short note before transcription helps with names and terms: "Interview with Dr. Amara Okafor about CRISPR off-target effects; host is Ben Li."

After transcription

Rename "Speaker 1 / Speaker 2" to real names, check quotes against the audio at their timestamps, then copy the text or download a TXT file.

Frequently asked questions

Do all routes label speakers?

No. Use GPT-4o Transcribe Diarize for speaker labels and timestamps.

Which file formats are supported?

MP3, MP4, MPEG, MPGA, M4A, WAV and WebM up to 25 MB.

Updated 2026-10-04. A focused OpenAI transcription workspace for recordings, interviews, meetings, and podcasts.