best settings for general use?

[viktorblack]

what would be the best settings for general use?

[francetoastVN]

I've done just some preliminary testing using the Balanced model, and trying the Aggressive, Balanced, and Conservative Sensitivity settings on the same video. The setting is all one apartment, so not noisy, not a lot of ambient sounds.

The Aggressive setting caught by far the most dialogue, but added a bunch of hallucinations during the long intro of the video, which was just cut scenes and music. No idea where some of the 'dialogue' came from.
But it did capture a phone call dialogue at the start, which was rather fuzzy sounding in the video.

The Conservative setting had almost no hallucinations during the long intro (except an odd "Subscribe to my Channel!" for some reason)
But...it completely missed the phone call dialogue! Just no subtitles at all.

The Balanced mode had a few hallucinations during the music intro (again, including "Subscribe to my Channel!' and "Follow us on Twitter") It caught the phone call dialogue.

I'm still comparing the overall dialogue-they are all different, so it's difficult to know which is most accurate (especially as my Japanese is very rudimentary), but for this video I think The Balanced mode did the best job.

I've just started exploring, but I really think it's a trial and error thing, based on the video content, settings, etc.
Plus this is just the most basic, quickest method to get subtitles-the Ensemble approach has some really interesting possibilities of really dialing in good subtitles-I'll explore them more and report back

Good luck!