First, our transcribers and editors produce a transcript of your video. Once we have a professionally written transcript, we put it through a voice recognition program that matches every word with an exact timestamp in milliseconds. We use this data to create downloadable subtitle and caption files in a variety of formats.
How good are these automatically produced subtitles & captions?
We think they are quite good. They are not quite as easy to read as if a human placed each subtitle, but they are more accurate and a lot cheaper.
How about some metrics?
We try to fit between 32 and 40 characters on a line for SRT and VTT, and between 28 and 30 on each of two or three lines for most other formats.
We try to show no more than 3.5 seconds of audio on screen at a time, and we don't let text hang around more than 2 seconds after it's been spoken.⬅ FAQ