Subtitling Pipeline

From source audio to translated subtitles

Transcription, alignment, and DeepL translation — orchestrated end to end and billed once at £0.025 per minute of source audio. Optional Premium AI Refinement layers register-faithful tone polish on top.

Three stages, one bundle

Each stage's project remains independently inspectable on your dashboard.

1

Transcribe

ASR-based transcription of the source audio in the source language.

  • Source language is forced — no auto-detect drift
  • Optional pause for transcript review before alignment
  • Transcription artifacts remain available on the linked project
2

Align

True forced alignment matches the transcript to the audio at phoneme level.

  • Word-level timing under VocaSync's alignment guarantees
  • 100% transcript coverage, no dropped tokens
  • Source-language alignment ready for translation
3

Translate

DeepL translates each cue while preserving the alignment timestamps verbatim.

  • Cue-by-cue translation; timestamps never shift
  • CPS-limit aware so output stays readable
  • Optional Premium AI Refinement (Claude Sonnet 4.6) for register
Bundled billing

One upfront charge. Auto-refunded if a stage fails.

The bundle covers all three stages. If a stage hits a terminal failure or the workflow goes idle for 30 days, the unconsumed portion is automatically refunded to your balance.

£0.025 / minute

Single rate covers transcription, alignment, and DeepL translation. 1-min minimum, 15-sec block rounding.

Auto-refund

Terminal failures release the unconsumed bundle back to your balance.

30-day expiry

Idle workflows auto-refund so reservations never sit on your balance forever.

Refinement à la carte

Optional Claude Sonnet 4.6 refinement, billed at +£0.006 / 1k translated chars.

Compare to running the stages independently: transcription (£0.01/min) + alignment (£0.02/min) + translation (£0.004/1k chars). The bundle simplifies billing and adds partial-failure protection.

Built for production

Common workflows where bundled subtitling earns its keep.

Anime fansubbing

JA→EN with anime-faithful register and CPS-bounded cues.

Drama localization

KO→EN, JA→ES, and other tonal pairs with drama-faithful refinement.

Podcast translation

EN→ES/FR/DE for global distribution with frame-accurate timing.

Conference subtitling

Multi-language tracks for talks, lectures, and keynotes.

Translate a feature episode end-to-end

Upload audio. Pick a target language. Walk away. The pipeline runs end to end and lands SRT, VTT, and JSON on your dashboard.