The youtube videos I want to use with Readlang rarely contain transcripts, so I need to generate a .mp3 file from the video and then upload the .mp3 to Readlang to generate the transcript.
Can this be done through Readlang directly so that the user doesn’t need to go through a 3rd party app to convert the video to a .mp3 file? Ideally, I would like to be able to import a youtube video directly into Readlang even if no transcript exists. If a transcript does exist, Readlang would use it, but if not, it would simply generate it.
Thank you!
Hi, not as far as I know, but there’s a quicker way, although it does involve an external app, and there are certain limits.
In these situations (almost daily) I use Turboscribe, whose free version alllows you up to three videos of up to thirty minutes per day. And it’s very quick.
You can even get speaker separation if you want. In this case it gives speaker labels and blank lines between speakers, but I use Readlang’s AI Polish - Other to get rid of these. I just have each speaker on a new line.
It won’t automatically sync of course, so I use Disable Sync and scroll manually. I could of course sync manually, but that seems to me a lot of effort for little gain.
Yes, I am familiar with Turboscribe and have found it useful myself. However, I often bump up against the 30 minute limit and don’t want to pay for yet another service. My suggestion was software implementation directly in Readlang:
Since the youtube import button already works in Readlang if a transcript exists, and Readlang also has the ability to generate a transcript from audio, why can’t Readlang just read the audio and generate the transcript automatically in cases where no transcript is available?
All the tools ARE already available in Readlang, it just would mean tying them together. If this were done, the import Youtube video would work for all videos, not just the ones with transcripts.
You can run whisperx locally on your machine. I transcribed a 30 hour audio book this way.