Auto generate timestamps not working as well as before

Hi there,

Sometimes the Generate timestamps button does a really good job of estimating where to put in timestamps and sometimes it’s way off.

Here is an example (Spanish) where the auto-generated timestamps are more or less perfect: Login - LingQ . At the moment, this lesson is private, so I imagine only LingQ staff can open it.

However, here is an example (Russian, also currently private) where the auto-timestamps are way off: Login - LingQ .

I had thought that the Russian example was faulty because I have added stréss áccents to the text (so I know upon which syllable the emphasis should go when reading aloud). However, this has not been a problem in the past.

Also, there are no stress accents added to the next example (German, private), whose audio is very clear. I was very surprised that the auto-timestamps did not find the correct solution for this: Login - LingQ

I hope that the tool can be restored to its former level of reliability. Doing the timestamps manually is too time-consuming to be worthwhile!

Thanks,

Steve

2 Likes

Yeah, I’m having this problem too, it’s literally arranging everything into really long segments to the extent that I can’t see the translations at the bottom of the Android app while in sentence mode. I was able to clean it up manually to some extent, but it would only allow for so many paragraph divisions.

I think we might be talking about different problems :slight_smile:

Mine relates specifically to a function LingQ has which can (did) split uploaded spoken audio fairly reliably into sentences, which can be played back in sentence view mode.

I’ve never encountered an upper limit to how many times one can use the Add paragraph spacing button in edit lesson mode. I hope you find a solution.

Thanks for reporting this. We will look into it and see what we can do to improve the feature.

1 Like

Thanks zoran!

This is working better at the time of this post. At least, the German example I gave in the original post now has more-or-less perfectly distributed breaks in the audio, thanks to the auto generate timestamps feature. I’ve just finished applying it to the whole course / book.

1 Like