They say it uses Adobe Sensei machine learning to match the pacing of human speech, and captions can be customised and embedded into the video itself or exported for upload to platforms like YouTube – which will be a big time and/or money saver for YouTube content creators.
The video above gives an overview of the highlights of the workflow, showing that the entire script can be easily viewed and changes made on the fly quite easily when mistakes or spelling errors need to be corrected. Premiere Pro’s Speech to Text includes support for 13 languages so far, with early access users apparently impressed by its accuracy, according to Adobe. While you can create these manually, Adobe says that the Speech to Text feature can save hours on your workflow, bringing the transcription time for a five-minute video down from an average of around 70 minutes to a little over 15.
The new feature for everybody, though, is the new Speech to Text engine that allows you to transcribe the audio in your videos, which Adobe says can result in higher discoverability in search engines as well as longer viewership.