Set primary folder (Portable version)

[dannbpz]

Hi, not sure what am I missing but i get this pop up when clicking on run, and the generate button doesn't do anything from my side

already run set_up_env.exe and have ffmpeg, 7zr and yt-dlp on the main folder

Thanks!

[cyberofficial]

Hello this error means that you didn't select the "transcribe_audio.exe" file. Click the button with the 3 dots " ... " and choose the transcribe_audio.exe program.

Setup set_up_env.exe will run automatically after you click run.

[dannbpz]

Hi, already selected transcribe_audio.exe but still have the same error

The CLI version seems fine but takes a lot of time to load even the small model.

Maybe need to add more settings than just --ram 4gb --device cpu ?

I have a Surface Pro 8 (i7-1185G7/16GB) and Python 3.10.9 installed

Thanks for the help!

[cyberofficial]

to be clear, you click generate on the ui and the message pops up?

[dannbpz]

ah no, the Generate button just doesn't do anything (idk why)

the message pops up when I click Run even after select the "transcribe_audio.exe" file.

[cyberofficial]

Could you try this build: https://github.com/cyberofficial/Synthalingua/releases/download/1.0.99995a441/Synthalingua_Wrapper_Aug_7_2024_Build_1.zip

Download this, and replace the files

[cyberofficial]

When using the wrapper,

When you do Click generate, click the unhide checkbox and see if it looks something like mine

[dannbpz]

Thanks for the build the gui now works! Now the generate button shows the script and run opens the cmd window

I was just wondering if is there a way to generate the Stream En translation text in smaller sentences like subs instead of paragraphs if I can only use base and small models

should I just ignore those user warnings btw? (FP16/FP32)

[cyberofficial]

should I just ignore those user warnings btw? (FP16/FP32)

Yea, I don't have a custom message for that error, so it just saying it's using a different system in place since your CPU is missing an instruction set for it, not that it's an issue, it's just using a different method that still works. Like a fall back. Not a thing you should worry about.

[cyberofficial]

I apologize for missing your other question

I was just wondering if is there a way to generate the Stream En translation text in smaller sentences like subs instead of paragraphs if I can only use base and small models

You can make smaller texts by choosing a smaller chunk size, https://github.com/cyberofficial/Synthalingua/wiki/Graphical-User-Interface-‐-Information#stream-chunk-size

Chunk size will download "x" amount of chunks from a stream. So you'll need to make lower chunk sizes for shorter inputs.