Audio transcription: turn a recording into text with speaker labels

Send an audio or video file URL and get back a written transcript, split by speaker. Use it for call recordings, interviews, podcasts and webinars you want to read or search.

What it tells you

• The full transcript of the recording.
• Who said what, as segments labelled by speaker.
• How many distinct speakers were detected.
• The language of the recording.
• How long the recording runs, in seconds and as a formatted time.
• The word count of the transcript, and the credits charged for the job.

What you need to provide

• An https URL pointing at the audio or video file to transcribe.
• Optionally, the approximate length of the file in minutes. This sizes the up-front credit hold.
• Optionally, turn speaker labels off if you only want plain text. They are on by default.

What it costs

6 credits per minute of audio. You are charged on the true duration of the recording. An up-front hold is placed based on the expected length you send, then settled against the real duration once the transcript completes. Live per-product costs are shown on the Billing page in your dashboard.

How to run it

• In the dashboard: Audio Transcription
• From your code: POST /v1/audio-transcription, then see the Developer Hub at app.1lookup.io/developer-hub.

Good to know

• Results are point-in-time. The transcript reflects the file as it was when the job ran, so if the file at that URL is later replaced or removed, the transcript you already have will not update itself.
• Because billing follows the real duration, a longer file than you estimated costs more than the hold, and a shorter one costs less.
• The file has to sit at an https URL that can be fetched directly.
• This is machine-produced text of speech. Expect a close record rather than a word-perfect one, especially with names, accents and background noise.
• Recordings often contain identifiable people and private conversations. Use them for legitimate business purposes only.

Updated on: 24/08/2026