Problems/ some questions

[coden965]

Im not sure where i went wrong i followed the installation instructions however these problems occured.

[cyberofficial]

Hello!

Few questions I have;

What Python version are you using?
Are you using nvidia or amd gpu?

are you running the script as admin or basic user? Please make sure to not run as admin if so.

[coden965]

Hello! Sadly I did just come off my pc, however I was not running it as an admin, I was using python 3.10 with an nvidia 3060

[cyberofficial]

Hello! Could you download this https://github.com/cyberofficial/Synthalingua/blob/a37c2e0771648132a31d75ed5e1ed60da5b9d3d4/requirements.txt file change and replace the requirements.txt file with this one? When you have a chance.

[coden965]

Hello, i just did this sorry for the long wait. However it is spitting out errors for torch.

[cyberofficial]

Okay,
since you have python installed

could you create a random folder anywhere and inside that folder run the command python -m venv .env

Then activate that environment with .env\Scripts\activate

With that try and run pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118 in this stand alone environment. If this doesn't work, then it'll narrow down the issue very much.

I'm also using python Python 3.10.8 to test and works for me, Python 3.10.9 should also work too as that's what's in my development environment

[coden965]

[cyberofficial]

You you have access to this page? https://download.pytorch.org/whl/cu118

[coden965]

That is what that page shows

[cyberofficial]

Okay, can you try this command?

nvcc --version

You should see something like this

[coden965]

[cyberofficial]

Could you try this in the terminal, in the sandboxed environment pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu121

[coden965]

Is this what you meant?

[cyberofficial]

You have a newer version of cuda installed so might need to bump the minimum version requirements

[coden965]

How would I do that?

[cyberofficial]

or might have to require a downgrade of cuda installed if the newest version isn't compatible

[cyberofficial]

pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu121

try this command to see if it installs

[coden965]

in the sandbox right?

[cyberofficial]

yes

[coden965]

[cyberofficial]

can you do python -V for me? and show the full the python version number

[coden965]

[cyberofficial]

needs to be capital V not lower v
you can exit that mode with quit()

[cyberofficial]

never mind is see the version number forgot the lower v shows it too,

[coden965]

[cyberofficial]

can you try pip install torch in the sandbox and see if that installs

[coden965]

[cyberofficial]

Falls on me for not specifying 32bit or 64bit, learning experience for me I guess 😅

make sure the x32 bit is fully removed, and try the original install method

[coden965]

Okay, it did run and download some thing however when im opening the setup.bash it gives this error.

[cyberofficial]

you'll need to run setup.bat* for windows. setup.bash for linux

[coden965]

Yeah i did run the setup.bat just unsure what to do after that (the livetranslation.bat doesnt work)

[cyberofficial]

did the setup.bat show success with out any errors?

[cyberofficial]

A good install would look like this

[coden965]

Yeah the setup.bat seemed to go in fine from what i saw. I've attached an image of what the livetranslation.bat give me. (yes the setup.bat looked the same as that)

[cyberofficial]

Ah, i see :) that means it's working

I need to update the example file, --non_english is a older command, you need to edit the livetranslation.bat file with new commands.

I've made a new UI here https://github.com/cyberofficial/Synthalingua/releases/tag/1.0.9997

Or direct download here Synthalingua_Wrapper_GUI-win-x64_Jan01_24-Build-6.zip this is the current UI.

Or visit the prelease page https://github.com/cyberofficial/Synthalingua/releases/tag/1.0.9998
and grab the latest build Synthalingua_Wrapper_GUI-win-x64_Feb9_24-Build-7.exe

and extract the main folder like so

[coden965]

Okay, ill go ahead and just use the wrapper gui what would i do with this?

[cyberofficial]

The main program will be called Synthalingua_Wrapper.exe

Search for the transcribe_audio.py

then choose microphone or HLS (default)
choose your gpu card's video ram size, stream language, then in stream url get a video stream (not video, needs to be live stream) click generate then run

You can check the video here #81 for for subtitle editing

[cyberofficial]

If you have have stuff like twitch prime, or twitch subs, or youtube premium https://github.com/cyberofficial/Synthalingua/tree/bleeding-under-work?tab=readme-ov-file#cookies you can check out how to use cookies to avoid ads

[coden965]

Okay i got everything however whenever i press generate and then run cmd pops up super fast which is pretty much unreadable

[cyberofficial]

can you show me the gui?

[coden965]

[coden965]

I did try it with a stream link however the samething happened

[cyberofficial]

make sure you have one these options enabled too

[coden965]

Yeah even with just one of them or both of them still no luck

[cyberofficial]

and you clicked HLS on the top left too?

[coden965]

i got it fixed

[cyberofficial]

Synthalingua_Wrapper_RxF40Vi8cY.mp4

You can also copy and paste each line from the command block into any command prompt too to debug it

[coden965]

The error i had i would assume is it being in far too many folders and i guess the program was just confused idk how to word it

[cyberofficial]

i see, windows has a char limit on name and folder lengths, something I can not fix, requires user tweaking, i can probably add a warning or something , but it varies per machine.

[coden965]

Well thank you for all the help you have given me everything seems to be working great now.

[cyberofficial]

Glad to hear it :D

If you have any more issues, please feel free to make a new issue.👍