Stereo Tool

After tesing many DSP tools, i came up using StereoTool Winamp Plugin for years (since v3) because of its sound quality and lots of customizing settings.
I wish: It could "feel" better the difference between slow-vs-fast music and between dynamic-vs-overcompressed music.
I mean: Soft settings (low amplification, slow attack) are OK for fast or overcompressed music, as it gives some dynamics. But it keeps slow or dynamic music soft. In contrast, hard settings (high amplification, fast attack) are OK for slow music, but adds more compression to fast and already overcompressed.
I think: AGC and Multiband attack-release speeds should depend to the differences between quiet and loud parts and auto set (maybe in some logarithmic way). For example, when a high-dynamics part pass through AGC, an auto <fast attack - slow release> setting, would give some push. In compressed parts, it should auto set to <slower attack - fast release>, to allow some dynamics.
I don't really know how DSP tools work, so maybe it is just another dumb wish, but I think this trick would equalize-normalize most of the songs in a music library. I also know that StereoTool is directed to a more loudness-signature sound, but I wish Hans should give a try.

Interesting comments.

I noticed on the default preset for example, some songs do sound slightly louder than others. I wonder if there is feedback through the entire DSP from the end result - to keep a steady average level more importantly than just peak control? I'd imagine there is.

It's a hard art tweaking an audio processor to give you the best results over all programme material. I still use an old 4 band limiter called MBL4 because it uses two time constants, so you can adjust it for two attack/release speeds in the multiband stage. I find dual time constants through allow you to work one for RMS and the other for peak control. The end is a more consistent mix. No where near as loud as Stereotool though.

Try my BASE presets:

viewtopic.php?f=10&t=3752
IMO it has a better loudness balance for old and new tracks than most other presets (because I use a relative low AGC output: - 12 dB RMS).

I have some other ideas that might help with tracks that have very little dynamics (some people have already seen a preview a few months ago, but the filter I wrote back then was not good enough yet).

About the volume difference: Do you determine that by listening or by measuring? And in case of measuring: Based on ITU-1770 level or RMS level?

It might help to turn ITU-1770 mode on in the AGC, that will base the volume measurement on apparent level for human ears (certain frequencies 'sound' louder) instead of on absolute RMS level.