I couldn't get it to work, I had to try a beta and portable version, but it was taking a long time and I couldn't get any results, so I'm using the virtual Collab instead and it's going great, I'm finally getting results, which makes the story go better.SE by default uses Whisper.cpp build which is CPU only ---no GPU dependency. I wonder if in your environment it is picking up a wrong build. That can happen I guees if you have installed (python version of) Whipser prior to SE.
Yes, a Collab, through drive, I don't know exactly what it is, I just know that some developers create codes to use Whisper easily without using their own resources, in my case I don't know anything about programming or codes it was easy to understand and use whisper+VAD and create very good subtitles and lots of lines of dialogue, so far the record for good lines is 1700 lines with good quality. It's not perfect but much better than the other methods I've used.virtual Collab?
convert my videos to audi (MP3) that weigh less than 200 MB even less, the movies, apart from the fact that uploading the file would take a long time, has limits as do other services such as auto sub or another, I recommend converting your videos to audio, then it will go like butter on bread.Just tried it. How do you get this to work with larger file sizes? Worked okay with a short clip, but when I try it for larger files, I get an error message.
I assume you mean the Google Colab Whisper WebUI GPU?
It is GPU. I don't have a GPU with lots of memory on this computer. Whisper converts the 2 hr files into something that is 2.9GB large and then goes on to find dialog. After running the audio it tells me I have number/220 and gives me a file of _ lines.Hello, I don't know why it gave you very few lines, I was testing with files of almost 3 hours, the movie DVDES-684 averaged almost 1700 lines and very good, you made sure to change the execution environment of the menu above to GPU, Well, most of the files I uploaded gave me incredible results, even with an old NSPS-068 movie with music and noise, half good dialogues.
View attachment 3117083
1. I am using a virtual environment.It is GPU. I don't have a GPU with lots of memory on this computer. Whisper converts the 2 hr files into something that is 2.9GB large and then goes on to find dialog. After running the audio it tells me I have number/220 and gives me a file of _ lines.
This is the audio I use: SGV-018
Can I get an untranslated Japanese file instead of a translated one? If I want to edit the mistakes the software makes I need to get a file in Japanese, not English.
Thanks.
I have to use Chrome. I can't upload files in Firefox. The wheel just 'keeps on turnin'.' Upload is slow, too.
1. I am using a virtual environment.
2. No need to use our computer resources
3. All work is done by Whisper's servers
4. Collab from the company Drive is used, I don't know what it is, but there are people who provide the necessary codes and repositories to carry out the translations.
5. All the movies I upload, I first convert them to MP3 with a converter, reducing the size up to 100 MB.
6. It's free up to a certain limit.
7.you can use another DRIVE account and use more Whisper time.
olab.research.google.com/drive/1CvvYPAFemIZdSOt9fhN541esSlZR7Ic6?usp=sharing#scrollTo=Aejy9Haxy81U
maybe 5 times, reset after a few hours. The trick is to use multiple gmails at the same time.What are the usage limits on Whisper Colab? Are they time limited (24 hrs) or do I have to wait several days to use it with GPU again? Not going to use Pro version.
I have achieved a new record for lines of text, before it was 1700 very good lines, now I have achieved more than 2600 lines, Whisper is incredible, at least for me... the movie in question was to test the reading ability of whisper, that a colleague wanted to know.
RCTD-459 Big Tits Mothers And Sisters Don_t Notice When Inserted Into Me (WHISPER+VAD)
View attachment 3119828View attachment 3119829