I thought I generally understood this until the other day when I was discussing this new headphone technology that Dolby has developed:
http://www.dolby.com/dolbyheadphone/
Basically, what it comes down to is a technology that allows virtual surround sound from two speakers.
Anyhow, the answer as to how Dolby does this is not what I'm looking for (I think it would be a little much to expect!). I just wanted to pose this hypothetical question:
If you are in an unlit anechoic chamber with a single sound source and your head is fixed, can you determine if the sound source is in front of you or behind you? And how?
The side to side part is easy.
Steve
Full Version: Direction of a sound source
It is possible with complex noises with high frequency content, because their spectrum is altered differently at the input of the ear, producing characteristing combing in the frequency range, according to the direction they come from.
It is impossible with sines. You can try it in full light, in a normal room, just sitting with the head free to move.
Put a speaker in front of a friend, another behind him. Then play a 1000 Hz sine in both, continuously, and set the volume to zero.
Now, turn up the volume (progressive analog volume control needed) for a very short time (1/4 of a second), on one speaker only. I bet your friend will never be able to ABX the front speaker from the back one
Important : don't play sine with a start/stop command, as the burst of frequencies occuring when the sine is suddenly interrupted allows to localize the source. The faint tic-tic-tic steps of a digital volume control, or a discrete analog one, may also allow to localize the source. It is absolutely necessary to have a pure sine set on and off progressively.
You can even try the test on yourself. You can let a sine play in one speaker, and try to imagine it comes from the other, it's very surprising when you turn your head in order to localize the speaker, you can get the feeling that the sound is rushing from where you imagined it came from, into the good speaker, as you move your head !
It is impossible with sines. You can try it in full light, in a normal room, just sitting with the head free to move.
Put a speaker in front of a friend, another behind him. Then play a 1000 Hz sine in both, continuously, and set the volume to zero.
Now, turn up the volume (progressive analog volume control needed) for a very short time (1/4 of a second), on one speaker only. I bet your friend will never be able to ABX the front speaker from the back one
Important : don't play sine with a start/stop command, as the burst of frequencies occuring when the sine is suddenly interrupted allows to localize the source. The faint tic-tic-tic steps of a digital volume control, or a discrete analog one, may also allow to localize the source. It is absolutely necessary to have a pure sine set on and off progressively.
You can even try the test on yourself. You can let a sine play in one speaker, and try to imagine it comes from the other, it's very surprising when you turn your head in order to localize the speaker, you can get the feeling that the sound is rushing from where you imagined it came from, into the good speaker, as you move your head !
QUOTE (Pio2001 @ Feb 27 2003 - 10:47 PM)
It is possible with complex noises with high frequency content, because their spectrum is altered differently at the input of the ear, producing characteristing combing in the frequency range, according to the direction they come from.
It is impossible with sines. You can try it in full light, in a normal room, just sitting with the head free to move.
Put a speaker in front of a friend, another behind him. Then play a 1000 Hz sine in both, continuously, and set the volume to zero.
Now, turn up the volume (progressive analog volume control needed) for a very short time (1/4 of a second), on one speaker only. I bet your friend will never be able to ABX the front speaker from the back one
Important : don't play sine with a start/stop command, as the burst of frequencies occuring when the sine is suddenly interrupted allows to localize the source. The faint tic-tic-tic steps of a digital volume control, or a discrete analog one, may also allow to localize the source. It is absolutely necessary to have a pure sine set on and off progressively.
You can even try the test on yourself. You can let a sine play in one speaker, and try to imagine it comes from the other, it's very surprising when you turn your head in order to localize the speaker, you can get the feeling that the sound is rushing from where you imagined it came from, into the good speaker, as you move your head !
It is impossible with sines. You can try it in full light, in a normal room, just sitting with the head free to move.
Put a speaker in front of a friend, another behind him. Then play a 1000 Hz sine in both, continuously, and set the volume to zero.
Now, turn up the volume (progressive analog volume control needed) for a very short time (1/4 of a second), on one speaker only. I bet your friend will never be able to ABX the front speaker from the back one
Important : don't play sine with a start/stop command, as the burst of frequencies occuring when the sine is suddenly interrupted allows to localize the source. The faint tic-tic-tic steps of a digital volume control, or a discrete analog one, may also allow to localize the source. It is absolutely necessary to have a pure sine set on and off progressively.
You can even try the test on yourself. You can let a sine play in one speaker, and try to imagine it comes from the other, it's very surprising when you turn your head in order to localize the speaker, you can get the feeling that the sound is rushing from where you imagined it came from, into the good speaker, as you move your head !
Recognizing directions of sound is a very difficult matter, and Dolby Labs
only mentioned things which are not showing the problems of systems
like Dolby headphones.
In a paper of "Production Partner" there is a full featured article about
a full featured headphone systems.
Direction recognition depends on the following things:
- level differences of signals on both ears => horizontal direction
- phase and time differences between both ears => horizontal direction
- colorization of wide band signal => horizontal + vertical direction
- change of the effects mentioned above when the head is moved => horizontal + vertical direction
- correlation between audible and visible events => front/back location
Also very important for a realistic sound reproduction:
- body vibrations of signals below 150 Hz for medium loudness (80 dB) and
below 500 Hz for higher loudness (110 dB).
Don't trust any information you find on the internet
when the Pros and CONS are not mentioned.
I'd swear I can tell, blindfolded, if a single digital click (one sample) comes from behind or in front of me...
But the ABX got me : I was still trying to remove the repeat option in ABX comparator, A being left, and B being right, that I already mistook a click coming from front for coming from behind
But the ABX got me : I was still trying to remove the repeat option in ABX comparator, A being left, and B being right, that I already mistook a click coming from front for coming from behind
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.