Subtitle Base workflow

Subtitle Base runs two services back to back:

1. Transcription — speech recognition turns your audio into a transcript.
2. Forced alignment — that transcript is aligned back against the audio for word- and segment-level timings.

You get SRT, WebVTT and JSON in the source language, with the same alignment guarantee as running the services individually: every word in the transcript appears in the output, in order, with a timing.

Use Subtitle Base when you are working monolingually — captions for the language you recorded in. Use Subtitling when you need those cues translated as well; it is the same two stages plus translation, under one charge.

Set autoApprove: false to pause after transcription so you can correct recognition mistakes before alignment runs. Your corrections become the input to the alignment stage, so fixing a misheard name here fixes it in the cues.

A paused workflow waits as long as you need and is never auto-refunded. See Subtitling for the full review flow, including how to approve over the API.

One upfront charge of £0.03 per minute of source audio, with a one-minute minimum and per-minute rounding. If a stage fails terminally, the stages that completed are charged at their standard rates and the rest is refunded automatically.

Source languages are limited to those the aligner supports, because both stages have to handle them. See Forced alignment for the current list.