I've been wondering... what is the rough cause of the 128kbps bass trouble in LAME 3.97b2?

Like, what makes this problem happen at low bitrates like 128kbps CBR/ABR and probably V4 VBR and lower... and not at higher-quality settings... ?

[edit: typo]

you know how it works here:

provide ABX results
upload a 30 second sample
state the settings you used

thanks

You beat me to it

NeroRenegade is talking about issues that are already known and which were discussed on the site for time to time since several months. Furthemore, there was already an attempt to fix it, without success (see here). And to finish, the "noise problems" were debated very recently (here, here, here and also here; plus several other local discussion you can find by using the search feature).
Do you really think that NeoRenegade has to bring proofs and samples just to have the right to ask a question about the cause of known issue?

I wasn't aware that it was a known issue. My apologies.

no and I wasn't aware of that issue. Has it been proven to exist "at higher-quality settings... ?" as well? Because that's what made me wonder...

There's a trumpet sample which produce audible noise at 320 kbps (consequence of a wild usage of short-blocs and insane bitrate with --vbr-new); there's also a clarinet one I've posted producing noise with VBR only and up to -V0 (this time, bitrate is unusually low => the problem should be different from the trumpet sample).

At least for samples like the trumpet sample there is hope that things will improve a lot in the near future, at least for higher bit rates.

Gabriel posted an experimental lame_attack version, and it yields transparent encodings for the trumpet sample to me for -b256 or similar abr usage.

I do not have so much hope for VBR usage, at least with current way of using VBR. Maybe VBR with the higher quality levels can be used more defensively kind of way abr and cbr are working (more of brute force, less intelligence that can be faulted) .