Rethinking Mid/Side Decoding
Most M/S decoders do exactly what they say. They decode. Left becomes Mid, Right becomes Side, and the result is technically correct, but often something feels missing.
Usually when you run old master through this, like "The Beatles" or "The Kinks", while they aren't technically real M/S Encodes, it sounds better as our ears are used to have natural stereo stage. Especially on headphones. Albums from the late 60s and early 70s were mixed with extreme stereo separation: drums hard left, guitar hard right, vocals somewhere in between - or worse, fully committed to one side.
With a modern decoder, the result is spatially interesting but vocally hollow. The center feels thin. The singer loses presence, as those hard-panned sides are collapsed onto them. The problem is structural.
In an aggressively panned mix, most of the vocal energy lives in the Side signal. M/S decoding recovers the width but can't recover what was never planned to be in the Mid to begin with.
What I changed
Zenteek's M/S Decoder now runs a parallel processing path.
Alongside the decoded signal, the original dry signal is collapsed to mono, passed through a bandpass, around where the body and vocals are usually sitting, and blended back into the output at a carefully calibrated ratio.
The result is a Mid/Side Decoder that sounds like it has a memory of where the music came from.
Mid Presence
I also added a dedicated knob named "Body" that appears when M/S Decoding is active, replacing the "Width" control.
When you leave it at zero it is the default M/S Decoder you already know. When you turn it up, it gives you direct control over the intensity of the parallel blend to full presence recovery.
It's like having an extra volume control for the center channel, allowing you to push the vocals forward. As you turn it up, the singer comes back.
However, that alone often is not enough, because some engineers decided to place the singer a bit off-center.
Panning Vocals
Music was mixed for loudspeakers in a living room where natural cross-feed allowed for some kind of staging, because our brain blends it all for us. On a headphone, this does not work as intended and Zenteek now fixes both issues.
First of all you can activate Zenteek's Crossfeed DSP alongside M/S Decode. That gives you a more natural presentation already. Second, when M/S Decoder is active you also gain a new pan-control under the "Body" knob.
You can use it to move the blended center energy - the body/vocals - back to the middle.
Once set it usually works for all tracks on that album.
Disclaimer
Please note that this is not a blue print. Some recordings may not work. I thought there was more to gain from this decoder and put you in control. Let me know outliers when you test this and what you perceived as wrong. I am happy to test those myself and tweak the DSP some more.