Is the site using Large model when using Whisper AI? It has the best accuracy but generates transcripts longer.
Alternatively you could add all the videos you want to download to one list and periodically download them with yt-dl.
I actually thought it was faster than other sites ive used
Does the automatic transcription not work for Swahili? It seems like the underlying whisper service does support Swahili.
@George_Pig We will look into it and check why it doesnât work for Swahili and what can be done. Thanks for reporting!
Hi George, it does indeed work for Swahili on our end. Iâve messaged you in Slack about this so we can troubleshoot it together.
Any idea when we can expect the end of lesson chunking for all imports guys? Thanks.
@SeoulMate - As it stands right now, there is no lesson splitting for Whisper imports and YouTube imports of 90 minutes or less. All other imports have had the split threshold increased to 6000 words. Anything over 6000 words will be split into 6000 word parts.
Thanks Mark. Do you know if the lesson splitting will eventually be lifted for other imports as well?
The 6000 word limit adjustment should be applied to all imports except the ones listed above. Which other imports do you mean?
Hi Mark,
Thanks for the reply.
In your OP, you said lesson splitting was no longer necessary due to performance improvement and that this would soon apply to all imports.
Thatâs why I am slightly confused by this new 6000 words limit. I thought there would be no limit when importing ebooks for instance.
So, is this 6000 words limit permanent or will the restriction be lifted at some point?
I was really hoping I could finally import books without having chapters split at all.
I would appreciate if you could clarify this. Thanks.
The reality is that we have still found that performance degrades with long length lessons. We have tried to find a limit that still performs well. At the same time we are continuing to work on the performance of longer lessons and have some ideas on things that may enable us to no longer have to split.
With regard to books, Iâm not sure importing the whole book as one lesson is desirable. I would think some ability to have it split into chapters would be preferable. This is also something we are going to be working on.
But, for now, you can assume that the status quo is how things are going to work for the next little while. Nothing else is imminent.
I resubscribed because of whisper integration, thank you for your work
This is really awesome! Iâm using Whisper now to add subtitles to a Youtube audiobook that I otherwise wouldnât have been able to import. I resubscribed just for this. Thank you for improving the site and making this change!
Hi Mark,
I use this new transcription feature on my computer, and itâs great. But can I also use it on either my iPhone or iPad and, if so, how? The import function on these mobile devices seems much more limited than on the computer. When I try to import via them all I get is this:
Am I missing something obvious?
Thanks
That feature isnât available on the mobile apps. Functionality on all platforms may not be the same. We are unlikely to add this on mobile since the handling of audio files on mobile is awkward. One interesting thing to note is the Scan feature shown in your image. It allows you to take a picture of any content with your deviceâs phone and have the text imported as a lesson on LingQ. Not something you can do on the web app of course.
Thatâs a pity, but thank you for this comprehensive explanation.
Thank you so much for the transcription integration! This has saved me so much time. I had been previously using an external transcription service, then having to put it into LingQ with the audio files. This is especially useful, as I use a combination of book learning (with audio only files) and app learning (i.e. LingQ and other apps) for my intermediate Japanese.
What doesnât work? What are we looking at here?