News at Princeton

Thursday, April 17, 2014

Web Stories

Video: 'Engineering 3-D sound'

To view the multimedia features on this page, you will need to download the latest version of Flash Player and/or enable JavaScript.

Engineering professor Edgar Choueiri has developed a way to play true 3-D sound recordings over regular loudspeakers. Read more.

Video Closed Captions

Edgar Choueiri: My name is
Edgar Choueiri, and I'm a

professor of mechanical and
aerospace engineering here at

Princton University.

Recently, I have been involved
academically in acoustics.

3-D sound is for you to sit in
front some speakers, ideally

two loud speakers only, and be
able to play an orchestra or

ensemble or a band, and be able
to locate unambiguously

the location of each
sound source.

You have a choir, you
can point to every

person in the choir.

If you have a symphony
orchestra, you can see the

viola coming from here, or
the base coming all the

way from the right.

In other words, a 3-D audio
image of that event.

Quite often people ask me, how
is that different from

surround sound?

Surround sound is just a way
to get you enveloped in the

sound, which works fine if
you're watching an action

movie with a lot of explosions
around you.

It doesn't attempt to
reconstruct 3-D audio image.

For example, if you want to
portray somebody walking to

the listener and whispering in
their ear, you can never do

that with surround sound,
because the sound is always at

the speakers.

But with 3-D audio, I can get a
fly to circle your head from

two loudspeakers.

[fly buzzing]

The first goal of my work in
this lab is to fundamentally

understand how humans
locate sound in 3-D.

If you are in a concert hall,
and somebody makes a sound

from the right part of the
concert hall, even if your

eyes are closed, you can tell
that that person is standing

all the way to the
right-hand side.

The brain can tell, because it
receives three kinds of cues.

The first kind of cue is simply
the time it takes for

the sound to get to your right
ear is slightly shorter than

the time it takes for the sound
to reach your left ear.

That time difference between the
two ears is enough for the

brain to analyze very quickly
and realize that the sound

must be coming from the right,
and not from the left.

The second kind of cue is that
sound, as it travels from the

source, when it hits
your right ear, it

has a certain level.

By the time it hits your left
ear, it has a lower level.

So to record in 3-D for
loudspeakers, all we need is

two microphones.

If the microphones are placed
inside the ears of a dummy

head, the microphones record the
correct 3-D cues needed for

humans to hear in 3-D.

The trick is in the playback.

The reason you don't hear in 3-D
through normal speakers is

that the left speaker contains
the 3-D cues for your left ear,

and the right speaker
for your right ear.

But these cues get corrupted
when your left ear hears the

right speaker, and your right
ear hears the left speaker.

This is called cross talk.

And without cancelling the cross
talk, the cues get mixed

up and your brain won't get the
information it needs to

hear in 3-D.

We need to now take what is on
the right channel, make it go

to the right ear, in the left
channel, make it go out to the

left ear, essentially
putting a wall

between the two speakers.

By sending negative and positive
pressure waves from

each speaker, my filter does
that so that the left speaker

sound never reaches the right
ear and vice versa.

Back To Top