One of my personal quirks, or curses for some people, is that my physics background has me question everything! Including things that I think I know…
[ The finest Astronomer I ever knew had an uncanny ability to take any question and turn it upside down and inside out and discover the most interesting things to investigate ever deeper. God rest his soul… ]
So this morning I fired up my Lavry DA-11 to listen through with my headphones. I’ve had this unit a number of years, and although the front panel very prominently displays some glowing LEDs and controls on either side for something called PiC, I have basically ignored that all this time… until today… what is PiC?
I went to Dan’s website to download a copy of the manual for my DA-11, and he talks in marketing-speak about the virtues of PiC for headphone listening. You know the problem — stereo tracks were mixed in a room with loudspeakers, intended for playback on a loudspeaker system. The listening room mixes the L/R channels in the air and your ears hear what they should, which is some room mixed combination of the L/R channels. But headphones disrupt that room mixing and present too much stereo separation to each individual ear. PiC, like my HdphX, attempts to overcome that problem.
But PiC was touted as a patent pending process. And whenever anyone says that, it becomes an excuse to obfuscate. In part, you must obfuscate during the patent review process. But a patent is a full disclosure on how things work. And once it’s awarded the whole world can dump your obfuscation into the trash bin and skip the marketing hype.
So I went searching the USPTO and found to my delight – both for myself and for Dan’s happiness – that this wasn’t just an empty marketing claim. Indeed, Dan had been awarded a USPTO Patent #8,306,232 for Digital Audio Stereo Imager. Congratulations Dan!!
And so now I also get to skip the marketing hype and see what he is doing.
It turns out that PiC is a really very simple process. Just scale each channel and blend a little bit to the opposite side. No Haas Effect stuff at all. Just a direct addition of some fraction of the opposite channel.
So how well does it work? Offhand, it sounds pretty good. I disabled my HdphX inside the running Crescendo and narrowed the image field a bit with his PiC controls, and I’m listening to some Hans Zimmer right now as I write this post. The stereo imaging seems pretty good.
The math is very simple. And it isn’t very difficult for my HdphX system either, which really does attempt some Haas Effect for added realism.
In Dan’s PiC processing, he rewrites L/R channels as composed of some true side plus center sound: , and
So by scaling and adding,
In other words, the new left channel has the original center information, plus a scaled back bit of the left side and some fraction of the right side, and vice versa. Presumably fraction k is near to unity, so kl is almost the same as the original left side. And (1-k) is small for adding a small amount of right side.
For HdphX the process forms the side channel as , and then delays that side channel by about 600 usec before filtering it with a lowpass + bandpass combination to represent the shading effects of the head, and some surface wave propagation at 5 kHz, to produce S’. Then the new L/R channels are formed as:
In this case, the scaling k is about -6 dB, or 1/2. The presence of that subtracted attenuated, filtered, delayed, signal on each side will produce some mild amount of comb filtering. Its depth should be very small, but not zero. Most musical power is in the bass region, which dominates the middle, M, channel, and comparatively little power resides in the side, S, channel. So any comb filtering should be almost unnoticeable. (but, is it?)
Both systems work okay…. Apparently, for most stereo mixes we don’t really need the Haas Effect to gain realism. It’s all fake anyway, so adding more fakery doesn’t seem to make it any more or less real.
But how would one judge the difference between PiC and HdphX? What tests could be devised to have a showdown competition between them? I’m still thinking about that one… (I personally think Haas Effect processing is the correct way to go, but does it really make any difference?)
Years ago, in the first versions of Crescendo, I worried that each ear would need different corrections – that most of us would have asymmetrical hearing. Well, we do, but that isn’t very important after all because of brain mixing. We never hear in only one ear. And as long as some sound enters the opposite ear, the brain will mix them together. So you may as well adopt a compromise Crescendo tuning for your hearing and feed both ears with the same adjusted sound.
[ I also worried that we wouldn’t be able to use room loudspeakers with Crescendo because that room air mixing would deconstruct the separate corrections applied to each ear… But you really can. ]
But I didn’t know that back then. And so I treated each ear separately in Crescendo, and I invented my little HdphX to help us with the stereo separation problem and headphone listening while under Crescendo processing. I thought I had to completely forego any post processing for headphone treatment. And so I never allowed myself to experience Dan’s PiC processing in the DA-11.
Fast forward these 10+ years, and the situation is much more forgiving. Go ahead and treat yourself to some headphone processing. Both ears are getting the same level of Crescendo treatment now, so it doesn’t matter if you mix opposite channels together, post-Crescendo. (but don’t do any filtering post-Crescendo!)