Subtitle Edit - Whisper Faster and Czech Speaker is ignored

[Matt Kirby]

I have czech talkshowfile

there are only 2 person who speak. I want to transcript the audio to text.
I use Subtitle Edit 4.0.1. for that with "purfview's whisper faster"

Most time the moderator is ignored by Whisper Faster (large model v2 and version 153)

Whisper CPP works good but has other issues.

The interview is longer but here's a sample

https://drive.google.com/drive/folders/1RAYmLzKYwXjt_3C0xtDTks8ovdx4DzKY?usp=sharing

Are there any parameters that I can use so whisper faster will recognzize the man at the left?

[VoodooFX]

There is dedicated thread -> https://forum.doom9.org/showthread.php?p=1986365

Later I'll check your sample, meanwhile post what "--verbose=true" shows.

[Matt Kirby]

With "--verbose=true" nothing changed.

[VoodooFX]

Medium model transcribed both speakers.
Somewhere I read that it can happen when speakers have different loudness and that some custom audio normalization can help in such cases.

With "--verbose=true" nothing changed.
You need to look at the output in console or in SE's logs.
It's not needed anymore.