I found a machine translation for BKD-186, I cleaned it up a little bit and tried to interpret some of the machine dialog. However it may not be a real translation because I do not comprehend Japanese. In any case take pleasure in and let me determine what you believe.
TmpGuy said: Why device translate to Chinese? Eventually, the intention for the majority of Here's to translate to English. We now have so many Chinese subs since they're the ones generating subtitles, not a lot of Japanese, that have little require for them (Maybe Besides Listening to impaired men and women).
one. 1st I wrote a script that checked each and every subtitle in opposition to the database of R18.dev. I believe script ran for around 15 hrs to examine above 20k subtitles. The challenge here is always that actresses that debuted in the final 3 calendar year are only available inside the database with their Kanji name inside the r18 DB.
Otherwise, typical observe should be to rename them to .txt so they may be opened and viewed uncompressed.
These are bare bare minimum Whisper Substantial subs. VAD with a very very low threshold but a higher patience and best of, along with a repetition penalty.
three. To battle the final issue I parsed my comprehensive map tree in several AI to talk to if it noticed any duplicates which have this problem. Deepseek, Chatgpt couldn't contend with it, they just invented duplicate names. Claud.AI context window was far too small (about 3000 actress names).
one. Initial I wrote a script that checked each subtitle against the databases of R18.dev. I believe that script ran for around fifteen hours to examine in excess of 20k subtitles. The challenge in this article is actresses that debuted in the final three year are only accessible from the databases with their Kanji identify during the r18 DB.
Not even remotely as simple as common commandline whisper. This seems to be a thing that pretty much Everyone is crashing so I think about website it can get preset.
Following some forwards and backwards with one sentences with ChatGPT I realized Marie Konishi is alternating involving calling the exact same person a naughty boy and dirty previous gentleman, which built Claude go nuts
The fundamental move of Whisper is it makes an attempt to detect single speaking segments of an individual speaker, and makes an attempt to determine the more than likely transcription of that dialogue chunk. It is actually supplying probabilities to a complete bunch of various possibilities, and after that choosing the one particular chunk that seems to be the most probably.
mayortommy claimed: precisely what is The easiest way to manage subtitles when two characters from the movie are speaking concurrently or in a short time after one another?
It might not totally be a true translation considering the fact that I don't know Japanese or Chinese. (Guess I should really often include this disclaimer!). Anyway love and let me really know what you're thinking that.
That is a tender Tale about a 'maiden' with no sexual knowledge along with a nobleman from your Heian time period, who's pulled on the current time through a Area-time warp.
one. 1st I wrote a script that checked every single subtitle in opposition to the databases of R18.dev. I are convinced script ran for around 15 several hours to check more than 20k subtitles. The problem in this article is that actresses that debuted in the final 3 yr are only obtainable in the databases with their Kanji name inside the r18 DB.