Link to Opusrug, it's a common OPUS encoder, no extra options, only some settings changed (celt file in the link). It's the one in the folder "opus 1.3.1 - 64 dirty by me" (.exe file for Windows), I didn't compile myself, and I didn't check if it has AVX2, feel free to check it.
Opusrug uses lowest possible bitrate while having highest sound quality, good for high quality melodic music. I don't guarantee high quality, you should use for music at least 64 kbps (though it's optimized for 56-80 kbps), good flac from a complete master, not singles if you want max quality.
DON'T expect M4A or Shazam audio quality as it has few psychoacoustics I edited only bass and bandwidth allocation.
I edited the celtencoder.c in a Linux virtual machine executing opusdemo without listening to the track. And I tested mostly with house and music from South Italy.
For compressing speech there are good encoders like Exhale or Gxlame that are more tested.
IMO this would be much easier to understand if it was released as a fork of the original repository, to be able to see clearly what was changed in the code. And easier to maintain when original codebase updates, too.
The problem is I don't have yet 1.3.1 - 64 master, someone compiled it with MABS and I don't think opus versions are retrocompatible.
This would be unfortunately my last update, as I don't plan changing the quality, add psychoacoustic. Also it isn't meant for low bitrates like 46 kbps.
But I posted the same.
Yes, I didn't mentioned streaming uses, but there could be and I don't know if could have a better preset for 46-48 music without re encoding. But as magicgoose said I think is difficult as it an old version of Opus.
try 39 kbps VBR speech if you want to encode singles with this encoder.
Talking about high quality at 144 kbps CVBR detect automatically it can have a similar quality to opusenc.exe 1.3.1-71 at 194 kbps CVBR speech and you want less bitrate re-encoding mp3s.
I'm a little confused... the only file you changed was celt_encoder.c? Can we see a diff? Maybe you want to submit this upstream if it has major improvements.
I can't use powershell, I uninstalled Visual Studio IDE and most website have limit of 2000 characters. So I don't know how to do, I'm not a programmer.
I re did tests
It's useful maybe for --bitrate 98 --vbr --speech
but nobody has ABX my encoder.
When I optimized it was for 56-80 kbps and melodic.