/transcribe — Transcribe Audio & Video

Convert audio and video to text The /transcribe command turns an audio recording — or the audio track of a video file — into a timestamped text transcript. It runs on Whisper v3 and automatically d...

Convert audio and video to text

The /transcribe command turns an audio recording — or the audio track of a video file — into a timestamped text transcript. It runs on Whisper v3 and automatically detects the spoken language.

How to use

Attach an audio or video file to the chat, or paste a direct link to one
Type /transcribe
Send — the transcript arrives as a normal reply, split into timestamped segments

/transcribe [with an audio or video file attached]
/transcribe https://example.com/interview.mp3

Attach one file per command.

Supported files

Input	Details
Audio	MP3, WAV, M4A, WEBM, OGG, AAC, FLAC
Video	MP4 and other video files — the audio track is transcribed
URL	A direct link to an audio or video file

Uploads are capped at 500 MB per file, and recordings up to 4 hours fit in a single run: whole lectures, podcast episodes and long interviews.

What you get back

Timestamped segments — the transcript comes back in chunks with timing information, so you can locate the exact moment something was said
Automatic language detection — Whisper v3 covers ~99 languages; no language setting needed
Text in the chat — copy it, translate it, or feed it into /summarize for the key points

The transcript is one continuous voice track. It does not label which speaker said what — for speaker-separated meeting notes, a dedicated diarization tool fits better.

Credits

Transcription is billed per minute of actual audio, measured from the returned timestamps:

Recording length	Approx. cost
Up to ~90 seconds	3 credits (minimum)
10 minutes	~20 credits
1 hour	~120 credits

Minimum charge 3 credits, about 2 credits per minute beyond that
Up to 4 hours of audio per run
Failed transcriptions are refunded automatically
Requires a paid plan (Pro, Ultra or Max) or a team seat — see plans

Use cases

Meeting notes — transcribe recorded meetings into searchable text
Interviews — convert recordings into quotable, citable text
Podcasts — make episodes searchable and easy to repurpose
Voice memos — turn quick recordings into written notes
Lectures — get study notes from a recorded class
Video content — pull the spoken text out of a video file

Tips for best results

Clean audio transcribes accurately; heavy background noise, music and crosstalk introduce errors
Higher-quality recordings produce better transcripts
Check quotes against the audio before publishing them

/summarize — condense a long transcript into the key points
/tts — turn text back into natural-sounding audio
Keep — transcripts and recordings stay attached to your conversation
Prefer typing? The microphone button in the chat input dictates straight into the composer at 1 credit per minute — no /transcribe needed.

More in Slash Commands

See Krater pricing or contact support.