Recent content by hury

I think it's a mixture between both the model and VAD, but mostly the model. Technically VAD should actually help though. I got around the same amount of hallucinations with VAD and large-v2 with "regular" whisper as well. Hallucination is a somewhat frequent subject in the whisper github, and...

Thanks, worked perfectly after I turned on "Multipart Encodings":keringat::D I have previously tried matching files to a scene in the same order as your picture (on a couple of movies here and there over several months), but it didn't help at the time. As an experiment, I created a scene for...

They've made it much easier. They've implemented the ability to import JAV from different databases. You have to choose which database and then write the movie ID you want to import. So you no longer have to create the scenes manually :)

Anyone else using xbvr (web interface) + Deovr / Heresphere (quest 2) with JAV VR? I'm wondering if some else has/had my problem: JAV VR videos usually comes split into several parts. I will the add them to xbvr as one movie with multiple scenes/files. Deovr / Heresphere registers it as one...

https://www.javlibrary.com/en/?v=javli7zzrq This is a subtitle that is a combination of a OCR ripped chinese sub (which was actually very good) and Whisper transcription. They complimented each other well, and again, the chinese sub was very good compared to regular ones. I haven't done anymore...

That sucks. Was about to suggest using Anaconda for separate environments to avoid things like this, but I see it's already mentioned in your OP. I linked to https://github.com/guillaumekln/faster-whisper in my post, but as mentioned, I run it through Faster Whisper Webui. I haven't tested it...

Just a little PSA. I've been using Faster Whisper through Faster Whisper Webui (https://huggingface.co/aadnk) for some days now, and I really recommend trying it out. I have a 3080 and I've been using large-v2 as model, and with faster whisper I'm cutting from 20 to 60+ minutes on tasks, using...

Have you tested with Japanese signs also/only (like なめて ちんぽ やりまん ぬれて)? I wonder if that would work better since it's transcribing, or is this flying over my head, and English is the way to go? Not OP. I worked a lot with ChatGPT on it, and after a lot issues etc. I got it to work perfectly...

Noticed this as well. I have generally just changed "comfortable" to "good", as that's how I have interpreted it. Thought with Japanese > Chinese > English, something would be lost in translation. But reading your comment, I spent a few minutes checking/listening things out in one JAV-movie...