[Newb] How to get full libs to use this whisper-standalone-win completely offline?

[BanQuanChung]

Hello.
I'm brand new and know nothing about coding. Just looking for ways to transcribe mp3 files into text and I have found this to try.
(Main language: Vietnamese)

1. Please show me the way to download all what is needed to use this completely offline.
2. Is it necessary to constantly keep the files up-to-date so that this machine works more and more precisely with transcribing?

Many thanks to Purfview and any others contributing to this code.

[Purfview]

First, "whisper-standalone-win" is not some program to be used per se, it's just the repo name, so, you can just read it.

I'm brand new and know nothing about coding. Just looking for ways to transcribe mp3 files into text and I have found this to try.

Looks like you found the right repo as there no coding involved to do what you want by using standalone executables.

How to get full libs...

You don't need to get any "full libs", whatever that means.

1. Please show me the way to download all what is needed to use this completely offline.

Just download a binary you want to use, then [auto]download models you need and it will be completely offline.
All that you can find on the main page.

2. Is it necessary to constantly keep the files up-to-date so that this machine works more and more precisely with transcribing?

Not necessary, but updates are for something better, not worse.

[BanQuanChung]

I got it, Purf.

I have tried faster whisper with the following commands,
faster-whisper-xxl.exe %file_list% -pp -o source --batch_recursive --check_files --standard -f srt -m small --language Vietnamese --device cpu
but the precision for the Vietnamese language transcription isn't as high as what I expect. And it took like 38min to finish for a 15min mp3 file. Quite a long time!
(Initially, I tried using auto captions function in CapCut Video Editing Soft. It's more precise and much quicker. But it requires internet, while I am looking for offline methods.)

Could you please help me with enhancing the precision in Vietnamese transcribing using this faster whisper?

[Purfview]

the precision for the Vietnamese language transcription isn't as high as what I expect.

Of course it isn't with the small model, use large-v2 model for the best precision.

And it took like 38min to finish for a 15min mp3 file. Quite a long time!

Buy better computer with CUDA GPU.

PS
Please don't post here a nonsense generated by GPT.