It’s great that there are so many lessons at Lingq (I’m working on Spanish) and really marvellous that so many of them have audio.
However, it is frustrating, every time, to find that audio does not sync to the SENTENCE VIEW within the lessons.
I’ve mentioned this before. I’ve said how great it would be if only it was ALWAYS used.
But, honestly, frankly, the way it is currently mostly just sucks. Syncing audio to sentence view is a great feature and yet almost nobody uses it when posting study material? Why not? It is because the interface is confusing and poorly explained.
So, PLEASE, can’t you sort this out?
LingQ is great but fixing this would make it instantly twice as good.
Also, please hurry. I’m trying to be patient but if this isn’t fixed soon I’ll probably leave LingQ.
Are you seeing it not sync on the app? The web?
One thing I had seen is I’ve generated the TTS audio on the lesson on the app, if I upload audio later to the lesson, the app will still use the TTS audio. You have to clear the audio cache on the app.
Lately adding audio to lessons has been working for me starting from a clean slate…import lesson. Upload audio file in the edit lesson. Then click “generate timestamps”. Everything working well in that manner.
Maybe if you can give more details on how exactly you are creating your lessons? Maybe there’s a step missing? Or maybe it’s an issue like I had above with audio already cached on the device that was autogenerated prior to me uploading my own audio?
I honestly didn’t even know this was a feature. I remember seeing Steve talking about it in one his videos but I never knew it was implemented.
I am sorry things aren’t working out for you. I just tried out the feature on a couple of Youtube videos and it’s working as expected for Russian. It even worked correctly for Esperanto videos. I am going to assume your source videos don’t have their subtitles aligned. There isn’t much they can do without using a lot of compute power to fix / evaluate imports.
Watch the source video and see if the subtitles are wrong / off there as well.
I am sorry that this function does perform poorly. But I find the sentence mode inefficient and it’s better to stick to the original passage mode. In the sentence mode, it usually takes me way too much time (probably 1.5-2X) than the passage mode. Yes, I do have a better comprehension, but I don’t think it worth the time cost. So probably stay with the passage mode as long as you can understand around 80% of the meaning, you are learning and much faster. I always find what I don’t understand at first will be OK naturally in a few weeks.
But I do hope they can improve their sentence mode, as I’d like to appreciate those famous books, which tend to be way too advanced for me, but meanwhile they are just so attractive and interesting.
If I create my own, sure, it works, as you say.
But so many of those lessons generously shared by others just don’t work properly.
I think it would help if you provided more guidance for people sharing material with audio.
So that it’s right there, on the page, when you are going through the process.
For example, the “generate timestamps” thing. That should happen by default, in my opinion, because it works so well and makes the audio so much more useful.
It could be that it doesn’t autogenerate because it doesn’t always work great. It does if I have nice clean single person audio like that last things I’ve imported. In the past I have had trouble with it not syncing correctly for certain sources. Usually probably where more voices and noise may be in the background. So many times for those, I’ll just prefer to have it use the tts voice and I won’t generate timestamps.
Certain languages might also not behave well.
Not sure about the instructions–they may have all this suggested in their help area on editing lessons. If not, it would be a helpful addition.
At least when I import text+audio the timestamps are automatically generated, for new imports no manual intervention is needed.
Older lessons require the owner / librarian to manually click “generate timestamps” in the edit lesson window.
My suggestion is to try to contact either the user who originally shared the content you are interested in and ask them to generate the timestamps, or contact a LingQ librarian for Spanish, they can trigger it as well.
If anyone is curious the software LingQ uses to provide this functionality is called aeneas: aeneas: automagically synchronize audio and text. It works well for many languages, I have extensively tested Chinese and Romanian although some manual adjustments might be needed (±1 second).
Although I try to put in time stamps (because I agree, that sentence view feature is great), creating those time stamps is VERY time consuming. Automated ‘generate timestamps’ does not work for Korean at all.
The good news is that people can edit time stamps themselves if they care enough to so. And have the time. I’ve done that for some lessons that others have imported into LingQ years ago, before time-stamping sentences was available.
I was able to edit time stamps in the Korean library as a regular user before I became a librarian. Is that available for other languages?
@Pippington I imagine some of the content is older and perhaps the ‘generate timestamps’ was not available back then. This could be a lot of content! So I think it’s a historical thing why a lot of content doesn’t have timestamps. That being said, the timestamp function doesn’t always work great anyways. With YouTube video imports, it’s usually pretty good. But I was trying to do it today for a podcast and it just was useless. The podcast had an advertisement at the start (not mentioned in the transcript) and the transcript had a summary/introduction (not in the audio), so it was just useless. I tried to manually change it, but with the lag and bugs and idk what else not working, I just rage quit it.
Yes, as a regular user you can edit the timestamps on a sentence by sentence basis. This is pretty great as it has the preview too, so it’s super easy to tweak.
Granted it takes awhile if you’re needing to do a lot of editing, but definitely awesome when the auto version doesn’t work as well.
I couldn’t agree more. For me it’s valuable to work through a lesson one sentence at a time while listening to the audio, possibly several times.
However, I’ve found a lot of great shared lessons with audio was poorly sync or that seems never to have been synched at all.
I do know how to create my own lessons and correctly synch an accompanying audio file (although this took a lot of trial and error because I could not find any well-defined material on how best to do this). Now, what to do about other contributors’ shared lessons? I can synch the audio for any such lesson one sentence at a time, but that has no effect on the lesson as it exists on Lingq’s servers, or does it? It would seem to me that the editing I am doing is for a local copy of some of the lesson content. Alternatively I can download the lesson text and accompanying audio, then create my own personal version of the lesson, but this is a time sink. What’s the purpose of having to recreate what should already be readily available in a usable form?
None of these issues seems to be clear from documentation and I get the feeling that no one who can address these issues is doing anything about them. I would like to leave Lingq because of this issue, but what is the alternative. Language reactor only works on laptops and desktops and, further, does not have a feature to add a separate audio file to a text file.
I wonder if I will get any constructive feed back on this.
That’s what I found. I stopped bothering. Now I just go with the Nice TTS Lady while in Sentence View.