PDA

View Full Version : Audio mixing- perception of sound direction


brodski
9th December 2006, 03:18 PM
At the risk of asking a stupid question...
Is it possible to fool a human ear (or rather brain) into hearing "3d" sound with just two speakers?

As I understand it we can ascertain the direction of sound based on the difference between what our left and right ears hear. With the closer ear hearing a louder sound, slightly before the more distant ear hears a quieter, similar sound.

Given this , would it be possible to use just two channels of audio in such a was as to convince someone that they where hearing just bone sound, which was coming from, say, over their left shoulder? or directly behind them?

Is this already being done?

ChrisC
9th December 2006, 03:29 PM
With the right pair of headphones, binaural recordings can sound like they come from 360 degrees. I don't know exactly why it works, but it's a really cool effect. Some headphones work better than others, and it's not neccesarilly about how good or expensive they are. I have an ancient pair of Sennheiser 414's that work great for that effect. It's completely lost on stereo speakers. I know there's some samples floating around. One that I remember well was someone shaking a match box and lighting one of the matches.

Myriad
9th December 2006, 03:48 PM
I don't know much about high-end audio, but it's clear that there are more variables to manipulate than merely the relative left-right volume of each source track. Most obvious would be phase changes (time delays). Volume or timing could be made frequency-dependent. Subtle frequency-dependent reverb effects could probably mimic the natural reflection of directional sound from the surfaces of the outer ear.

Digital audio processing makes this all relatively easy. But some producers were getting the same kinds of effects decades ago, by the simple expedient of recording from microphones located in the ear canals of an accurate model head.

Respectfully,
Myriad

jon
9th December 2006, 04:21 PM
some TV's and hifis have a 'virtual surround' feature - I think they do largely mess with the phase, and it can make the sounds appear to move back/forwards as well as left/right. Nothing like being in a cinema with a good soundsystem, but fun all the same :)

Dog Boots
9th December 2006, 08:19 PM
Yes, it works and it is being done - but not very well on speakers, headphones is the way to go with this.

It can be done both artificially (calculated) and by recording with a dummy head (or a real/your own head!). Some people record with their own head, by sticking two miniature microphones inside their ear canals and record a regular stereo signal that way.

It works because our brain knows what to expect to hear from different directions. It knows how our outer ears shape the incoming sound depending on the direction, and it works by measuring the difference in timing between the ears. If a sound reaches the left ear half a milisecond before it hits the right ear, the sound will be located by your ear as coming from the left.

There are technologies out there that can do it artificially. A Danish company a few years back had an editor available for free download, so people could play around with their 3D-engine, that was meant to be implemented in games and such (I don't know if this technology is implemented in games yet, but it sure can't be far away now).

D'oh - just found it (Thanks, Google!)

Check this out: http://www.am3d.com/526FABBC-7EB1-45B7-8DB5-344342D9A894.W5Doc

The application is called Diesel Studio AM:3D and it's GREAT fun to play around with - a 4-view editor, just like in 3D-modelling programs, and then a chopper sound loaded in by default (you can load in any sound you want). Fly that chopper around your head and pickup your jaw afterwards! Also be sure to check out some of the demos.

(These calculations also stem from recordings made with a dummy head originally, btw.)

The human psycho acoustic capability is incredible. For example: have you ever wondered, when recording with a camcorder in a regular, not too acoustically dampened room, why there's always so much room-sound (reverberation) and the recording - much more than it sounded like when you were there, making the recording? This is because your ears filtered it out, when you were there in person. Your ears can distinguish between the direct sound of the person talking, and the reflections coming from the room, so it filters out the reflection, and you perceive a much more dry (less reverberant) sound image. When listening to the recording (from the camcorder f.inst.) your ears can no longer make that distinction, because all the sound comes out of the same speaker - the direct sound from the voice of the person speaking, and the room reflections.

pipelineaudio
9th December 2006, 11:29 PM
Always keep in mind that we dont yet have a recording system which can even remotely, slightly, more than a passing resemblance-ly accurately recreate the sound of a source

Take the most expensive dummy heads you can find, a marshall 900, and a 4X12 stack

Record this

Play it back blind thru some monitors and tell me you cant pick real vs recorded every time

Thing
11th December 2006, 01:55 AM
These people claim to be able to do it:

http://www.isvr.soton.ac.uk/FDAG/vap/

They have downloads so you can try it at home, but you need your speakers to be closer than they would be during regular stereo listening.

Mr. Scott
11th December 2006, 08:38 AM
At the risk of asking a stupid question...
Is it possible to fool a human ear (or rather brain) into hearing "3d" sound with just two speakers?

Is this already being done?

It's a great question. The prefix "stereo" has nothing to do with the number 2. It means "solid" or, if you prefer, 3D.

Two speakers work only if the listener is equidistant to each speaker. Move closer to one or the other, and the sound stage moves and compresses towards that speaker.

To reproduce not just the sound but the 3D* sound wave front, you can record with two omnidirectional microphones and play them back into speakers located proportionately to where the microphones were. (To do it perfectly, you need a plane of an infinite number of microphones recorded into an infinite number of channels to a plane of an infinite number of speakers.)

To do it adequately for listeners who don't need to be exactly front and center, you can do it with five mikes/channels/speakers. This was discovered in the 1940s and commercialized in the 50s Cinerama movie process (drawing below). It actually had 7 channel sound -- five behind the screen and two for surround effects.

Here's one place on the web you can read more on Cinerama Sound (http://www.cineramaadventure.com/theworks-sound.htm).

The 7-channel Cinerama sound system inspired the 2-channel stereo system we are now familiar with.

*Strictly speaking, it's not 3D sound. The ears are most perceptive of the front 180 degree left/right dimension. We have very weak perception of front to back, up and down, or near and far. We use other cues for those dimensions, although it's been found that the complex shape of the outer ear makes resonant cavities that help distinguish up and down. It should be not too hard to simulate up/down in an audio reproduction system. We also tend to move our heads unconsciously to help us locate the sources of sounds. That fact works against the headphone method.

http://forums.randi.org/imagehosting/6736457d7b38cf4c7.gif

69dodge
11th December 2006, 09:19 AM
I remember trying out a digital piano one time, a Korg I think it was, first listening to its built-in speakers and then using headphones. With the headphones on, the sound still seemed to come from in front of me and not from the headphones. The effect was so realistic that I actually took the headphones off my ears, keeping them plugged into the jack, just to make sure that the sound I was hearing wasn't coming from the speakers. It wasn't.

alfaniner
11th December 2006, 09:23 AM
I still have a cassette I bought about 20 years ago, called Holophonics. It is basically a bunch of sound effects recorded in the manner stated above (mics placed on a head model). It included things like a hair dryer, a paper bag being placed over the "head", and a bee buzzing in your ear. These were three of the freakiest examples -- the bee had me ripping the headphones off and rubbing my ear.

brodski
11th December 2006, 09:36 AM
These people claim to be able to do it:

http://www.isvr.soton.ac.uk/FDAG/vap/

They have downloads so you can try it at home, but you need your speakers to be closer than they would be during regular stereo listening.

cool thanks, I'll have a play arround.

vIQleS
12th December 2006, 03:56 PM
I vaguely remember discussing this during my sound and lighting course...

It can be done with frequencies. It has to do with the shape of the ear...

I can't remember any details but I think 5kHz had something to do with something. (front and back i think)

Brian Jackson
13th December 2006, 02:42 PM
Phase, micro-delay and EQ are the key components to creating this effect. Roland had a 3D system years ago, as did, I believe, Lexicon. Cheaper systems are simply stereo image wideners/compressors that could be varied over time. An example is the Behrringer Edison, one of which I own.

A good recorded example of 3D stereo imaging is Roger Waters' CD "Amused To Death."

scotth
13th December 2006, 03:02 PM
Phase, micro-delay and EQ are the key components to creating this effect. Roland had a 3D system years ago, as did, I believe, Lexicon. Cheaper systems are simply stereo image wideners/compressors that could be varied over time. An example is the Behrringer Edison, one of which I own.

A good recorded example of 3D stereo imaging is Roger Waters' CD "Amused To Death."

Late to the party and beat to the punch....

Brian is exactly correct. The Amused to Death album does this amazingly well. It is processed with a system called Q Sound, but that is just a brand name covering a collection of components that Brian has already enumerated.

Something that is critical to make this work properly is near perfect symmetry of the listening space, with the listener sitting astride the line of left/right symmetry.