Automatically align a transcript
We bang on a lot about our amazing speech recognition, now in 28 languages. Quite rightly: with well recorded audio, it’s a massive time saver. We’ve had people write in to say that they’re seeing savings of up to 80%.
But we feel like our Auto Transcribe is getting all the glory. We’re just as proud of how we can import a transcript, with Auto Align.
Let’s back up a bit. In order to create captions, first you need a transcript. Second, you need to time that transcript out against the video. When you do this, you’ll want to put caption breaks and line breaks in sensible places, respecting natural pauses in the video, who’s speaking, grammatical rules, and the video’s edit points.
If your video has already been transcribed, then it still needs to be timed out against the video. We offer a few methods to import transcripts in CaptionHub, but Auto Align is almost always the fastest. Here’s how it works:
- Create a project in CaptionHub, as you normally would
- Take a plain text (sorry, no Word files yet) verbatim transcript
- When creating your original captions, choose Import Transcript, then choose Auto Align
- That’s it, you’re done!
What’s going on behind the scenes is that we’re using speech recognition to align your transcript. We’re basically telling the system: ‘Look, here’s a perfect transcript. Analyse the audio of the video to work out where it goes. And, while you’re at it, put caption breaks in sensible places.’
The net result is that the speech recognition has to work much less hard, and the results are almost always astonishingly good.