Stereo Tool

Hi,

I've been playing with StereoTool for a while, and as I said elsewhere, the tool's ability to manipulate phase seperately from the width is something I had not seen and lead to many hours of figuring out how to take advantage of this (like removing QSound processing).

Anyway, I have a suggestion and it's based on a system that was developed and used (to the ire of many people) in the late 60's that dealt with the transition of mono to stereo as a primary medium. I've encountered this system a few times on LP's, and even on CD reissues of LP's; and it got my attention both as an obscure form of what I call "audio cancer"; methods that did more harm than good, and as something that can be done in stereo phase.

The basic method of downmixing, which I'm sure has been around since the dawn of stereo production, has been to combine the left and right channels. However, this does not account for audio that is panned dead-center, or in other words, audio that is exactly the same (in phase) in both the left and right channels. What happens is, during a simple downmix, all in-phase audio builds-up in it's amplitude. So, for example, if you have something panned dead-center at -6db, during a simple downmix, this audio winds up max amplitude; if you had audio panned left/right at -6db, this audio would remain at -6db in the downmix.

The way this was avoided way back when was; by my own theory, by spending the extra money to make a second mono mix with properly adjusted levels; or just horrible mixes that had everything panned hard left/right. However, a system was developed known as Haeco-CSG. Haeco-CSG was an attempt to create, what was called, mono compatible stereo. It did this by taking advantage of phase cancellation, and rotating the phase of a channel by a specific number of degrees so that center content would partially cancel out; thereby being reduced in amplitude. This allowed record companies to produce stereo-only LPs that, in theory, would fold down to mono when played back on a mono-turntable, or summed to mono by radio station that did not broadcast stereo. The resulting recordings folded down to mono perfectly; however, playing this "encoded" stereo, in stereo, sounded pretty rough. While it's not quite the "inside-your-head" result a full 180 degrees causes, it makes everything sound louder to one side and just, off, in general. Engineers started actually mixing through this system, leading to horrible releases. Thankfully, it died a few years later.

I have however discovered using this creates a much better, balanced, downmix; vocals, which are usually panned center, no longer overpower the rest of the recording, leaving musical details and background vocals much clearer. No other plugin or editor I've seen uses this technique; despite the fact I think it's a good idea. In fact, I have an effect-chain programmed with a graphic panner and channel mixer for the times I want to go mono.

Now, the Haeco system was mostly used in what was known as the "+3 db" setting. While it still resulted in +3db of center build-up; the encoding was only 90 degrees of phase shift. This was entirely immune to accidental reversal of phase (polarity) in a channel; which could of happened due to mis-wiring of a turntable cartridge or somewhere else in the chain; the audio would still fold down to mono properly. The system did have a 0db build-up preset; which uses, from my calculations, a 120 degree phase shift. However, this setting was not used; if one of the channels had it's phase/polarity reversed, the resulting fold-down resulted in massive cancellation of audio.

However, in cases where the final output is already going to be a single channel; the risk of a polarity getting crossed is eliminated, and use of a 120 degree phase shift results in cleaner mono.

Now, StereoTool both helps and hinders this method in the way it currently works. While I enjoy the separate processing of phase and width; this operation can get in the way. As it stands, while StereoTool does create a better downmix thanks to the re-removal of phase before removing width; the center channel bulld up does not get resolved. In order to this, I have to run a graphic panner in the chain. Running it before SteroTool would yield no advantage; as the phase removal will undo not only phase difference that have occured in the mix; but the 120 degree shift added; so the logical point is putting it after. However, with the seperation of phase and panning adjustments; even adding a StereoTool afterwards does not work; the phase change will remain due to the processing of phase before width. I did test to see if there was some way with this processing that it mixed without suffering from center-build up. I bounced the same track to three tracks in multi-track and assigned each one a way of downmixing. A simple averaging, my method, and straight StereoTool. For the most part, the StereoTool sounded much like the simple downmix, with the exception of a bit of audio that got saved by it's phase processing. The elements were, for the most part, the same. While the backing track sounded fine, the vocals did seem to be a bit over the rest.

Based on my results, running the phase removal in StereoTool first benefits this process. There have been a few tracks where the straight phase shift before mixing yielded bad results with things canceling out more than they should; removing the phase relations solved this. However, due to the way the processing works, this is about all StereoTool has to do. Usually I run it from there in to a graphic panner and in to a standard channel mixer. Currently, to use SteroTool to properly fold this down, you have to leave the phase at 1, reduce the width, then roll the angle to 120 degrees; resolve it before you change the width and you're still not resolving center-channel.


I don't know how much, if any call there is to go mono these days; but I for a while have felt taking advantage of phase cancellation creates a mono mix closer to what it's supposed to sound like.


Thanks for the consideration.