Skip to content

Welcome to Subaligner’s Wiki!

Xi Bai edited this page Sep 25, 2020 · 3 revisions

Given an out-of-sync subtitle file along with a piece of audiovisual content carrying speeches described by it, Subaligner provides a one-stop solution on automatic subtitle synchronisation with a pretrained deep neural network and forced alignments. In essence, aligning subtitles is a two-stage process with a Bidirectional Long Short-Term Memory network trained upfront and Subaligner helps subtitlers not only in preprocessing raw subtitle materials (outcome from stenographers or STT workflow, etc.) but also in gaining quality control over their work within subtitle post-production. This solution also tolerates errors occurred in live subtitles which sometimes do not completely or correctly represent what people actually spoke in the companion audiovisual content.

For full documentation and API reference, please visit https://subaligner.readthedocs.io.

Clone this wiki locally