Stereo Magic with no Pan (?)

[rhythmboy]

Hey all, learned a nifty trick at an AES workshop at Swinburne Uni I attended last week. The speaker was Ron Streicher from the AES in America, a recording legend and foremost expert on stereo. He's co-authored a book called the 'New Stereo Handbook" with acoustics legend F. Alton Everst who wrote "The Master Handbook of Acoustics".

Amongst other things Ron spoke about the human perception of stereo and how we localise sounds in two main ways:

- at low and mid frequencies, it's the time delay between sound arriving at one ear before the other. The earlier sound arrives at (eg) the left ear, the more we sense it coming from from the left. If there is no time delay we perceive it being in the centre.

- at high frequencies, it's phase difference and the 'shadow' our head causes as sound passes around it from one ear to the other.

In the case of the first issue, Ron gave a demo that I attempt to replicate here, because when I heard it I was amazed at the impact of it for such a small and simple technique. In essence, he exploited the time delay issue to shift sound between two speakers.

This only really works if you sit in the sweet spot in the middle of your speakers.

First up, a male voice, ripped from a Mix Magazine bonus CD. Mono:

[mp3=200,20,0,center]http://www.soundpunk.com/downloads/home/RBstuff/malevoice0delay.mp3[/mp3]

Next, the same thing, with the right channel nudged 1 msec later:

[mp3=200,20,0,center]http://www.soundpunk.com/downloads/home/RBstuff/malevoice1msdelay.mp3[/mp3]

Now the right channel nudged 2 msec later:

[mp3=200,20,0,center]http://www.soundpunk.com/downloads/home/RBstuff/malevoice2msdelay.mp3[/mp3]

Now the right channel nudged 5 msec later:

[mp3=200,20,0,center]http://www.soundpunk.com/downloads/home/RBstuff/malevoice5msdelay.mp3[/mp3]

Notice how the voice seems to pan to the left? Now move your head in front of the right speaker - it's as loud as it ever was.

Finally, the right channel nudged 5msec later, but the left channel lowered by 3dB:

[mp3=200,20,0,center]http://www.soundpunk.com/downloads/home/RBstuff/malevoice-3db5msdelay.mp3[/mp3]

Bingo it sounds even in both speakers again.

Think about it next time you reach for pan... 

[Spectrum]

Great concept, and great clips. Nice work, RB.

I've racked my brains in the past trying to get what makes sounds travel around the stage. And you're right, it ain't all in the pan, so to speak.

Those examples are quite startling. And I'd not have guessed the effect would be so apparent. On a sidenote, it also reveals the 'damage' comb filtering does as each millisecond-delay example has a completely different tonal signature when compared to the original.

I've only ever duplicated/split synths/guitars/cymbals before (never vocals) to give 'em width (watching out for mono compatibility) and had never before (until now) spotted the panning effect (until the delay becomes so obvious to be able to immediately pick which channel is arriving first.) That demo is quite remarkable.

And continuing Ron Streicher's idea, a gentle (6dB/Octave) low-pass filter on the 'other' channel to dull/muffle the high frequencies might work well to emulate "the 'shadow' our head causes as sound passes around it from one ear to the other".

But then I wonder, what makes regular volume panning so effective if it's unnatural to how sound works in the "real world".

And when panning fully left (or right), the sound actually comes from that side of the room, so just by the location of the speaker with regards to the listener's head, it naturally forms the time-delay and shadow-effect phenomenon that's being raised here.

One limitation, however, to the time-delay or phase-shift concept is the size of our heads, literally.

With sound travelling at around 344metres/second, or 34.4centimetres/millisecond, and the human head spanning less than a standard ruler (30cm) from ear-to-ear, the time difference between the right and left channels, even if panned to the extremes, will never exceed 1ms.

Still, it is quite a revealing experiment, and has got me thinking about stuff.

[rhythmboy]

Oh why did I see such a fascinating reply so late at night  :bang:

Now I have to say something...

The comb filtering issue definitely has to be considered. Admittedly the voice does sound thinner and a bit hollow. Also with low-freq sounds the potential for phase cancellation is significant, so user beware, and don't expect great mono compatibility with the delay technique. However the effect could be creatively exploited too  I can imagine string pads, sweeping swirly sounds, guitars, anything already harmonically rich working really well.

The low pass filter would indeed create a subtle effect and help increase stereo perception of top end sounds like hats and tambourines - sounds we often pan to the side anyway. Interesting you say 6dB/octave because in Ron's demo he pushed it out to a 10msec delay - on my nearfields at the desk too long and you hear slap (like you said, the delay is affected by the size of one's head and proximity to the sound source). But in the auditorium through the little PA it was dramatic - and he pointed to a 6dB difference in perceived level. He boosted the delayed channel by 6dB and it was perfectly centred again. There you go, 6dB is the magic number for sound waves in many ways...

(sidenote for n00bs: pan is effectively a voltage-dividing potentiometer - in simple terms, like two volume knobs in one, it changes the relative level of each channel. As you pan left, the level in the right channel goes down and vice versa. This makes it appear to come from one side or the other. The pot also lets us control how much the level changes - if it's only a little, the pan is closer to the centre)

I think what makes pan effective is that we are starting off with a significantly different amplitude (loudness) of signal arriving at each ear, from a very focussed and predefined space (the speakers) which is then assisted by time delay and shadowing. The sources are more focussed than sounds in the real world, coming from all around us at once.

Yes pan works, but notice how sounds seem more extreme in the stereo field when you add delay too?

Stephen St. Croix from Mix magazine (The Fast Lane column) once wrote a thought -provoking article on the use of reverb to enhance stereo as well - or more to the point, how we often destroy stereo image with reverb (and pan too for that matter). As briefly as I can...

Stereo perception is also assisted by reverberant space - we also use reflections in a room to help decide the position of a sound. It will seem more direct in the direction it is coming from, and more reverberant on the other side of you head.

So never pan the reverb channel to the same position as the dry channel. Try panning the reverb to the opposite side you've panned the dry sound. If my tambourine is 50% left pan, send it to mono reverb and pan it 50% right. Or even more extreme, if the tamb is 80% left, try 20% or 80% right in the verb. The effect can be quite dramatic. I do this all the time now, when appropriate.

Also try using short and subtle decays but plenty of early reflection in the reverb - the mild slap effect increases the stereo effect.

Whew... heavy but very useful - kids even you don't get the theory try out the techniques - they really do work 

[ant]

This is a very common trick to push live guitars wide of the vocal in a live band situation.

An intersting way to take it further is to layer multiple live takes, all with different but complimentary filtering. For example if you have three takes, then each take is filtered to one third of the spectrum you want the vocal to use. Then apply slightly different delays to each one. This will give you a very processed sound particularly good for backing vocals. Think Christina Aguillera. More takes / bands, the better. There were over 130 vocal channels used for 'Dirrty' in Pro Tools!!!!

[VUMod]

Yes cool idea mate - delay and filter multitracks to create space and width. Must try him out more - I often only think to do one or the other

Wonder if they did it to Enya as well? Those 80-track vocal sessions...

I sure hope others read this thread - there's some real 'secrets of the trade' here.

PS this is really RB here, just logged in under my school name 

[groovewell]

absolute quality stuff here guys, i just gotta get home from uni and try some of this stuff out. Awesome