Top watch jav sub Secrets
Top watch jav sub Secrets
Blog Article
@Imscully - you, together with @AgarabjeM, brought me one of the most pleasure for my work on this thread = I have every one of the respect for your function and i am glad you failed to throw in the towel!
Yeah, I checked this collection out a handful of several years in the past which video with Kawai Anri was the ideal of them.
I've deleted A lot of them just via encoding all files to UTF-8 without bom then checking If your filesize is similar. But certainly if someone puts an ad in there, the filesize differs...
mei2 mentioned: Would any person know a Doing the job approach to batch translate these files? The scripy deeplv4.py has stopped working for me --I believe a change in DeepL UI has induced it.
Japanese is among the much better languages for this by common transcription and translation metrics, but dirtiness and pronouns might be a Recurrent trouble to work around.
You are not just translating with whisper, you happen to be typically transcribing the audio to text, that's the aspect that requires usually, the interpretation is essentially an right after imagined for whisper so evaluating to deepl line by line is evaluating apples to oranges, Except if deepl has an choice to translate directly from audio that I do not find out about.
My new x265 rips: mkv=best quality + 10bit(not blurry like most x265 rip) and mp4=really smaller but good high-quality(the usual blurriness of x265)
Regarding duplicates, I had a look at the content material but did not see any variance, I will need to consider a more in-depth search afterwards.
Makkdom explained: I've run into 2 problems with the hooked up sub file. It's incorrect spacing during the notation for the timing with the dialogue traces. I was capable of correct that Along with the "substitute" operate on Wordpad. But the numbering of each line of dialogue is likewise off. The 1st line is accurately labeled as one, but then it jumps to 65 with Each and every line remaining consecutively numbered from there.
Some are thoroughly unedited, and the rest are just scarcely and have a huge amount of problems. These are not the identical good quality as my Slave Colour subs.
temperature: A measure of the amount of randomness goes into your transription and translaton course of action. This appears unintuitive in the beginning, but carrying out a bunch of Original conditiosn and looking at what comes out, and evaluating each of the probabilities and variety the best, creates improved outcomes.
As such, the majority of my assistance During this thread is going to be for maximizing the quality of the Medium models.
As typical, you'll find lines that I haven't translated, lines that I'm uncertain of (Particularly one that mentions New York), and a few uncomfortable phrasing, but I've carried out my best to help make the working experience a superb a single.
speedy scan of @Makkdom Model with @ericf Chinese Model, there are plenty of variations. But meh I'm as well lazy and Actually The entire vid is understandable more than enough Despite having inaccurate click here translation.