Subtitling workflow

The Subtitling workflow runs three services back to back:

1. Transcription — speech-to-text from your audio.
2. Forced alignment — precise word- and segment-level timings.
3. Translation — translated subtitles in your target language.

You upload audio once; the bundle produces timed, translated subtitles at the end.

Set autoApprove: false to pause after transcription and correct the transcript before alignment and translation proceed. Catching recognition errors here keeps the timings and the translation clean — your corrections become the input to the alignment stage, so a misheard name is fixed everywhere downstream.

From the dashboard, review and approve on the workflow page. Over the API it is three steps:

1. Subscribe to pipeline.awaiting_review — transcription is asynchronous, so this is what tells you the workflow is waiting.
2. GET /v1/workflows/{pipelineGroupUuid}/transcript to read it.
3. POST /v1/workflows/{pipelineGroupUuid}/approve to resume, optionally with an editedTranscript.

The bundle is priced upfront per minute of audio, at one rate covering all three stages. Register-aware translation is included rather than sold as an upgrade.

See pricing for current per-minute rates, and the workflows overview for how bundles compare.

Don’t need every stage? Two lighter bundles cover the common cases:

• Subtitle Base — transcription + alignment (no translation). £0.03 / min.
• Quick Translate — transcription + translation (no alignment). £0.06 / min.

Create subtitling pipelines programmatically with an API key and presigned uploads. Track progress with webhooks. See the API reference for the workflow endpoints.