Transcription
Upload a meeting, interview or call recording and get a transcript split by speaker, with timestamps. Then open it in chat to summarise it or ask questions.
The Transcript app turns audio and video recordings into text, split by speaker and stamped with times. Name the speakers, copy the text, and send the whole transcript to a chat in one click to get minutes, action items or answers to “what did we agree on pricing?”.
Transcribe a recording
- Open Transcript
Go to Apps and choose Transcript. The New transcript screen opens.
- Add the file
Drop an audio or video file onto the page, or select Browse files.
- Choose the model and language
Pick a TRANSCRIPTION MODEL (see below) and a LANGUAGE. Auto-detect works for most recordings; set the language yourself when the recording is short or the audio is poor.
- Open the transcript
When it is Done, the transcript opens with the model, duration, cost in credits and date at the top.
Supported files
| Kind | Formats |
|---|---|
| Audio | MP3, WAV, OGG, FLAC, AAC, M4A, OPUS |
| Video | MP4, MOV, AVI, MKV, WEBM |
Models
Choose from ElevenLabs Scribe v2, Gemini 3.5 Transcribe, GPT 4o Transcribe Diarize, GPT 4o Transcribe and OpenAI Whisper.
As the app itself puts it: larger models are more accurate but slower. If a transcript comes back with names or jargon wrong, try another model with Re-transcribe (below).
Work with the transcript
Name the speakers
Each turn is labelled with a speaker (spk:0, spk:1, …) and the time range. Select Speakers to rename them - spk:0 becomes Darian, and so on - so the transcript and anything you make from it reads naturally.
Copy it
Copy puts the full transcript on your clipboard.
Open in Chat
Under PROCESS WITH AI, Open in Chat starts a chat with the transcript loaded, so you can clean it up, summarise it or ask questions about it. For example:
Write the minutes of this standup: decisions, action items with owners,
and open questions. Keep it under 150 words.Re-transcribe
Under RE-TRANSCRIBE, choose a different Transcription model or Language and select Re-transcribe to run the recording again.
Tips for clean transcripts
- Record close to the speakers. A laptop microphone across a meeting room loses the quiet voices first.
- Keep one voice at a time where you can. Overlapping speech is the hardest part for any model.
- Set the language for short clips or recordings with a strong accent.
- Rename speakers before you summarise, so the minutes say who agreed to what.