Create documentation / Developers

Transcription and media review

Transcribe owned media, inspect results, and prepare timed captions or edit proposals through the API.

In this guide
A reviewed transcript or edit proposal to inspect before changing the timeline.

1. Choose the source asset

Upload or select media available to your account and project, then retain its asset ID. Transcription and review operations analyze that source.

Open API documentation → Endpoint reference and search transcriptions. Expand the operation to check the request fields. Follow API and integrations first if you have not created a key.

The transcription endpoint operates on a specific asset. Its details describe the request and asynchronous response.

2. Request and retrieve a transcription

Send POST /assets/{id}/transcriptions with the source asset ID and documented body. Retain the review identifier and poll its review endpoint until processing finishes.

Inspect the recognized words, timestamps, confidence information, and caption artifacts. Results can include SRT/VTT and comparison against an expected script when supplied.

Listen to the source while checking names, numbers, and technical terms. Resolve uncertain words before using the transcript as captions.

3. Review the media or finished render

Media reviews can inspect metadata, frames, audio, motion, transcripts, and captions. Bind a review to a completed render when checking a particular timeline revision and its caption timings.

Use the results to locate issues, then inspect those points in the video. Confirm that captions align with speech and that the output contains the intended picture and sound.

4. Translate or request an editing proposal

Create a speech-edit job from completed transcription reviews of the same source asset. Search speech-edits in the reference for its current request schema.

  • translate: returns English cues with original segment timing and uncertainty information.
  • edit-plan: proposes cuts, chapters, and an opening for review.

These results do not change the timeline automatically. Select an available text model and review pricing before submitting this billed work. Reuse the original idempotency key when retrying an identical request.

Speech-edit operations prepare text and proposals from completed source reviews. Apply approved changes in a separate editor step.

5. Preserve artifacts and apply approved edits

Download needed review artifacts through their authenticated content URLs. Current review artifacts expire after seven days; preserve the transcript and caption files your workflow needs within that period.

Review the proposals, apply approved edits, and export again. Inspect the new output before delivery. See AI editing and the timeline for the app workflow.