how are they doing it? any tips for a beginner?
Find an audio and let it play on repeat or have a playlist that loops through.
Did it for Chinese for 10,000 hours
1 Like
Playlists in Lingq work great. I get most of my content from Youtube. I send the link to Turboscribe.ai then download the resulting srt file and audio. I create a lesson in Lingq and add it to a playlist. Then I listen while I do other things like was dishes, mow the lawn, walk in the woods, and go running.
My playlists are always in infinite loop mode so the current list will keep playing. Every time I listen, I understand a bit more. This is especially effective after a day, week, or a month since I listened the last time.
Six new lessons for German listening today. BTW: I pay for both Youtube (to get rid of ads) and Turboscribe (because I use it that much). These are both great investments for my language learning. Used daily.
3 Likes
Is Turboscribe that much better than YouTube automated subtitles?
I don’t mind paying but it must slow down your workflow a bit.
Not sure what your workflow is direct from Youtube. Turboscribe can be used for free (three 30 minute transcriptions per day). So you can try it yourself and be the judge.
What Turboscribe gives me …
Turnboscribe gives me a time-stamped SRT file (many other formats are available). You can choose the level of accuracy, and I always go for the most accurate transcript that Turboscribe can provide, At the same time, the audio is captured. I can also choose the language to transcribe to, although I usually go for the original language. TS also manages my transcriptions so they continue to be available online as long as I desire.
If I can get all that faster and easier, let me know.
2 Likes
Well, I just import the video directly into LingQ via YouTube’s share functionality.
For Spanish at least, auto-generated subtitles are quite good.
Good to know. I would max out my limits way too quickly.
AFAIK there is no limit not on using YouTube-generated subtitles; it’s just text.
Thank you.
I have never used the Playlist feature on LingQ before. This should be a big help in the morning while just doing stuff.
2 Likes
True, but I don’t want them either.
In my experience, YouTube videos imported straight into LingQ break up the text along the lines displayed on screen. This means that a lot of sentences get broken up in ways that are unnatural and jarring to read, just because sentences are too long to display on screen at once. Is this alternative you mention here better at this? This alone would, to me, be worth paying for. That, and to make sure that when reading sentence by sentence the audio actually matches the beginnijg and end times of the sentence. (Which is off so often with the beginnner content on LingQ)
2 Likes
You can either manually modify the srt files (you can download them with yt-dlp, for example) or ask ChatGPT to do this for you.
For example, the srt file could look like this:
7
00:00:29,470 --> 00:00:31,290
知事。リーダーですね。
8
00:00:31,290 --> 00:00:34,750
で、この東京都のリーダーを決めるための、
9
00:00:34,750 --> 00:00:38,040
東京都の知事を決めるための選挙が
10
00:00:38,040 --> 00:00:40,320
今行われています。
11
00:00:40,320 --> 00:00:45,600
で、えっと7月7日に投票日ですね。投票があります。
It is relatively easy to see which lines end with an sentence ending (denoted with an 。), and which not. The only downside is, as can be seen in the last part, that you may have several sentences in one line, but usually only short ones.
The rest is deleting the parts you don’t need. I am not sure whether you have to change the numbering, as I haven’t tested that. So the first and the last part stay the same, but the three in between get merged (including using the end timestamp of the last one).
7
00:00:29,470 --> 00:00:31,290
知事。リーダーですね。
8
00:00:31,290 --> 00:00:40,320 🡐🡐 end time from line 4
で、この東京都のリーダーを決めるための、東京都の知事を決めるための選挙が今行われています。🡐🡐 merged lines 2-4
11
00:00:40,320 --> 00:00:45,600
で、えっと7月7日に投票日ですね。投票があります。
You can also use FasterWhisper XXL. It’s free. I am not sure whether the paid versions are that much better than free one, as they use the same algorithms in essence. My tests with shareware products in the past never gave me the impression that the paid products are better, but it might be different now.
The transcription, both auto-generated at YouTube or with other software depends on two factors, mainly. The amount of data provided (so languages used by a lot of people, like Spanish, have some benefits over less used languages) as well as the phonetic properties of the language, especially how strong the link between the written and the spoken language is (here, Spanish has some benefits again, as almost everything is written as it is spoken and vice versa).
1 Like
Yes, Turboscribe gives you proper sentences. I tried to import directly from YouTube into LingQ and via Turboscribe and prefer the way via Turboscribe, because I hate that I only get lines and not proper sentences the other way.
@Obsttorte It is great that there is a way around, but I don’t know enough about computing and I have no idea how I should implement what you suggest.
By the way, so far I do not pay for Turboscribe, because the free allowance is enough for me. I use it currently mainly for Ukrainian podcasts.
I am more concerned with transforming print to audio. I hate the voice that LingQ comes up with in Spanish
You don’t need to implement anything. FasterWhisperXXL is essentially an implementation. You start it via console with maybe some parameters. The same applies to yt-dlp. No programming or specific computing skills needed. That beeing said, instructions on how to use those or small scripts can be written for you by ChatGPT.
1 Like
Thank you. I will have a look at it.
Thanks so much, I will try this out!
1 Like
I take TV subtitles and read them in LingQ. Then I go back and watch the tv episode / movie the subs came from, with the subs attached. Then I manually add the amount of time the TV lasted to my audio listening amount. If I’m really struggling then I watch the tv episode / movie in English first before I do all this.
In languages I’m more advanced in, I just listen to podcasts, the news, or ebooks on YouTube when I’m out walking or doing the dishes or something.
Maybe they are leaving sound on all night as they sleep.
1 Like