Kraut wrote:sorata225 wrote:emk wrote:Kraut wrote:Though as it is it's still very useful and although its successors "Language Reactor" etc have more functions (well ..too many and constantly being reworked and ..ugly!!.) It's open source!
This is very nice. Does it have SRS integration?
The problem with Subs2SRS is that it's fantastic at the A levels, when everything is unfamiliar. But once you get past B1, the sheer effort required to find accurate, aligned subtitles and to make cards becomes way too much work for too few cards. I strongly suspect that I could benefit from Subs2SRS-like cards even now, especially for regional accents and extremely colloquial language. But it's just not worth it, in terms of time.
But tools like Lingo Player could be extremely promising if they could automatically make Subs2SRS-like cards—and if they integrated with an SRS system with more appropriate defaults than Anki. I haven't been paying any attention to this space for years—does anyone have a really good video-to-SRS pipeline integrated into a player these days?
I do still want to write up my ideas for good beginner course, with an emphasis on getting to solid intermediate listening skills sooner. Soon, I hope!
Are you aware that whisper can make pretty accurate subtitles? I was trying to find subs for Made in Abyss when I found a file with whisper generated ones. To my surprise, even for Japanese it was very accurate. And I'm quite sure it was not modified after generation. The reason being the show featured a conlang for the inhabitants of "Abyss" and as expected the AI tried to transcribe them in Japanese.
Yeah, there are also "freesubtitles", "voscribe, "downsub", "subtitles translator" ....
oh no I wouldn't compare those with the open-source whisper ASR from chatgpt. Cannot imagine others to be on the same level.