Did you get any results with the translation using Whisper,
@tim-o-tea ? Is there an easy way to do it from Windows? Thanks!
subtitles-support in XBVR/Heresphere:
for those of you who are using XBVR. Luckily subtitles now work with XBVR and with the newest Heresphere it also correctly works with multipart videos.just make sure you are using the newest versions of xbvr&heresphere. If subtitles are having the same filenames as the video files they are also automatically matched correctly.
subtitle generating with whisper:
I did some tests with whisper and while not being perfect its MUCH better than my results with pytranscriber(aka autosub).
It does generate a lot of lines for the actual dialogues than pytranscriber where often it created lines but it was just garbage.
One weakness is that dialogue that is "dirty" is not properly transcribed/translated, but it seems like this in an limitation in english only (in the sub thread, somebody did a test in another language, maybe there are some workaround for that).
Before converting anything you should always convert the videos to audio files (for GUI-Tools i recommend FFmpeg Batch AVConverter or just Audacity is just fine)
As for using whisper there are basically two ways. Run it locally or run it in the cloud (Google Colab).
If you run it in the cloud you have the advantage that you dont need good computer (GPU/CPU) as all the workload is done in the Google Cloud. Downside is that ressources are limited (some users are mentioning that they can only convert a certain number of files till they are blocked).
Here is the link the the cloud version:
https://colab.research.google.com/github/ANonEntity/WhisperWithVAD/blob/main/WhisperWithVAD.ipynb
Its important to note that its better to use Whisper with VAD because with the regular whisper it will create a lot of double lines unfortunately.
While a lot of people are using that cloud version, i had no success as it always created subtitles that werent for the whole video.
I had more success running it locally (keep in mind you need at least some kind of gaming GPU) using this version
I've found Whisper to be an incredible free tool for transcribing audio, so I've made my own WebUI which integrates directly with YT-DLP for direct YouTube transcripts, and allows for easy download...
github.com
The setup takes some time, but once set up & running you do everything on the User interface to its very easy. The PDF instructions on the link are also VERY detailed so it should not be a problem even for less techsavvy people.