Q: mspectraldynamics

[bklnlks]

i have been testing and playing around with mspectraldynamics, attempting to arrive at the cleanest (least artifact-prone) setting for spectral compression, and cannot seem to arrive at a settings free of what sounds a little bit like transient-smearing pre-ringing (otherwise notorious for phase-linear EQs); others may liken it to lossless compression artifacts. (i assumed the highest 'quality' along with the quickest 'resolution' would yield the cleanest results.) is this something inherent in the design that cannot be avoided, or could someone point me to what it is that i am doing wrong? thank you!

[pone]

Higher precision (resolution, lower attack/release etc.) will actually introduce more artefacts like this, you can also try to use a higher sample rate.. for things to be different (the spectral resolution will be smaller but the processing resolution will be higher).

[mljung]

i have been testing and playing around with mspectraldynamics, attempting to arrive at the cleanest (least artifact-prone) setting for spectral compression, and cannot seem to arrive at a settings free of what sounds a little bit like transient-smearing pre-ringing (otherwise notorious for phase-linear EQs); others may liken it to lossless compression artifacts. (i assumed the highest 'quality' along with the quickest 'resolution' would yield the cleanest results). is this something inherent in the design that cannot be avoided, or could someone point me to what is it i am doing wrong? thank you!

You're not alone - I've tried MSpectraldynamics many times but always end up using dynamic eq instead, as MSpectraldynamics seem to kill some important sound-subtleties one way or the other. It's making the processed sound a little life-less, hard to describe - but I have realized that many times where I A-B against dynamic eq (dealing with the same issues), MSpectraldynamics has to go...
Whether it could be a consequence of linear-phase filters - I reckon Vojtech can tell us

::
Mads

[pone]

Whether it could be a consequence of linear-phase filters

It is, a good demonstration is to use Mtransformer and compare the same settings in the level graph with a small buffer size and a large one. The larger the buffer, the higher the resolution, but the higher amount of artefacts (ringing) around transients.
I recall voxengo reducenoise having a minimal phase mode, however it has a 2 second plugin delay and has no side chain capabilities.

Try to use MSpectralDynamics on parts of a sound where transients don't matter.

[MeldaProduction]

Correct. Basically, it is spectral, hence linearphase and as such the preringing is and always will be an issue. It usually doesn't seem to matter (at least to me) when not overused though. Doing this minimum phase would be awesome, but if it would actually work, you'd still need a supercomputer for that