Compatibility of Subtitle Edit with Whisper's "large-v3" Model (January 2025)

[mamomo]

Hello,

I am trying to use Whisper's "large-v3" model (released in January 2025) with Subtitle Edit to improve the accuracy of auto-generated subtitles. However, I am encountering issues, and the model is not being recognized by the software.

What I’ve Tried:
I downloaded the pytorch_model.bin file for the "large-v3" model from Hugging Face.
Renamed it to large.bin and placed it in the folder where Subtitle Edit stores Whisper models.
Attempted to use Whisper for subtitle generation, but the program does not detect the model.
The Problem:
The model does not appear to load or is not recognized by Subtitle Edit. I am unsure if the software is compatible with this version of Whisper.

My Questions:
Is Subtitle Edit compatible with Whisper's "large-v3" model?
Are there any additional steps, configurations, or file formats required to make this model work?
If it is not compatible yet, is there a plan to add support for newer Whisper models in future releases?
Additional Information:
Subtitle Edit version: [Include your version here, e.g., 4.0.3]
Operating System: [Include your OS, e.g., Windows 11]
Error Messages: None, the model simply does not appear to load.
I would appreciate any guidance or advice on how to resolve this issue. Thank you in advance for your help!

Best regards,

[Purfview]

I am trying to use Whisper's "large-v3" model (released in January 2025)

This model was released in 2023... better question is where do you get this nonsense from? 😁

[mamomo]

Thank you for your response! 😊

You’re absolutely right—my mistake regarding the release date. The "large-v3" model I was referring to might have been released earlier than I thought (2023). I must have misunderstood or come across outdated information.

My main goal is simply to use a more recent version of the model than the one currently installed by default in Subtitle Edit. If you have any insights on how to achieve that or if it’s even possible, I’d greatly appreciate your help!

Thanks again for pointing that out and for any guidance you can offer! 😄

[darnn]

For a minimum of fuss, in the Audio to text window, select Purfview's Faster-Whisper XXL (CPP CuBLAS also works but is slower on my system). Download it if necessary. Now you'll be able to download the models it supports, including large-v3:

[mamomo]

Thank you so much for your guidance! 😊

I followed your advice and selected Purfview's Faster-Whisper XXL in the Audio to Text window. After downloading the necessary files, I was able to use the large-v3 model without any issues. It works perfectly, and I’m very impressed with the speed on my computer—it’s incredibly fast!

Thanks again for pointing me in the right direction. This solution has made a big difference for me. 😄

Best regards,