You really must have headphones on when you listen to this wonderful exercise in stereophony.
There’s more information on how it was made at toomanytribbles. You’ve gotta love your brain and the way it can translate phase and timing differences in sound into a spatial map, without you really having to think about it.
Some of that stuff sent shivers up my spine. I’m glad this isn’t Virtual Dentist!
Reminds me of the first time I listened to The Invisible Man by Queen. “Get out of my skull, Freddie Mercury!”
Haha, kinda spooky to think you can listen to a recording and still be able to get a lot of positioning of the voice. Unless of course he’s a ventriloquist.
http://www.amazon.com/Mist-3-D-Sound/dp/0671874756/
That’s a pretty good Stephen King short story done in the same fashion. Some of the effects sent shivers up my spine, even after the 3rd or 4th time listening to it.
Whoa! Sitting here in an otherwise empty office and listening to that with my eyes closed was just incredible. I second the comment about the dentist…
Waaay cool, but have to admit that I really hated his accent, and the way he all the time had to state out loud which side he was working on… right..onto the back…left etc.
That’s really neat. I’ve felt similar things in movie theaters, but the effects of this one are better.
More holophonic links:
http://holomatches.ytmnd.com/ <- demo using a box of matches
http://www.holophonic.ch/newsite/index.php <- I think these guys own the “Cetera” algorithm that produced both the matches demo and the Virtual Barber Shop. The intro sample on their page is pretty nifty.
Somebody needs to get the Firesign Theater one of these things.
Nothing new here. Binaural recording has been around a long time.
yeesh! the scissors! the scissors! nooooooooo! my ears!
LOL. this was cool.
Lost on me but then I wear hearing aids and have grosly mismatched hearing.
I experienced something like this at Disney World yeeears ago. It might even have been exactly this, because i remember a (virtual) barber and the very distinct sound of scissors cutting hair. I wish i could remember which park it was…probably MGM, as I believe it was in an attraction about movie magic. You went into a sound booth, put on headphones, and listened to the recording. There was also an exhibit on Foley artists in the same attraction. Damn, now I am going to have to go use my google-fu for the rest of the day trying to track this down!
I think it’s time for me to start a band called The Psychoacoustics.
Interesting Wikipedia article. I’m sure I remember Edgar Froese of Tangerine Dream releasing a binaural recording *mumpty-mumph* years ago, although it’s not mentioned. I remember it because I was quite into Phaedra at the time (having heard John Peel play it), and it was the first time I’d heard of the technique.
Darn! I saw “barber shop” and read “barber shop quartet.” I was all ready for close four-part harmony. . . .
I was just looking for binaural recording effects on YouTube yesterday, and here you have a great (if heavy-handed) example. I love the internet. It’d be great to see this sort of effect integrated into games. Second Life already has doppler effects, and the millisecond pauses and minute volume tweaking can’t be much more taxing on the CPU. Of course, I’m a sucker for immersion – I GIMP autostereograms from digital art, I pore over every new announcement of 3D consumer monitor technology, I own a Virtual Boy, for fnord’s sake.
YIPE! Beware the whisper! I flinched away from my left…
When it first started, I thought something had hit the outside of my window, which is about five feet to my right. I had to listen to that part again to make sure it was actually in the recording. It was freaky.
I can’t find any specific information on how they made it. There are really only 3 ways to do something like this. One is to stick microphone in somebody’s ear canals and record from them. The second is to use a dummy head with similar size, shape, and acoustic properties that has a microphone in each ear canal (Kemar is an example). The third is to use what is called a “head related transfer function”, which is a digital filter created from one of the other two techniques that reproduces the effects of the head. The advantage of the last technique is that all you have to do is record the sound and then tell a computer where to place it. The disadvantage is you need an HRTF for every location you want to have a sound come from.
I was about half-way through it when I realized my headphones were on backwards…”as I move to your right side” and I was like “hey! that’s my left—oh, duh…”
The whisper part nearly made me jump out of my seat, too.
The whole thing was WAY cool.
Great technology, but the best they could do was a… barber shop?
It’s boring enough having one’s hair cut, but at least at the end of it you have a new haircut. This is all the stupefying boredom of the droning hair stylist’s inane banter with none of the payoff.
Binaural recording can be astonishingly realistic. Some years ago I did a recording session using the very same Neumann dummy head mic system pictured in the Toomanytribbles blog entry.
The purpose of the session was to produce a recording for a museum exhibit on hearing. The dummy head was suspended from one of our big boom stands and the two clients walked around the head, speaking and playing various percussion instruments.
I was monitoring on headphones, and after a few minutes found that I had to avoid looking at the action in the studio because the discrepancy between what my eyes told me about the location of the sound source and what my ears told me about the same was making me nauseous.
Dummy heads like the Neumann, which model the shape of a human head fairly closely and have very realistic pinnae do the best job; they capture all of the acoustic effects- timing differences, shadowing, diffraction effects and even pinna transforms (yes, there’s evidence that the folds in your outer ears play a role in localizing high-frequency sounds)- that our heads and ears have on the pressure waves impinging on our eardrums.
There are two problems with this type of recording. First, binaural recording still doesn’t reproduce front-to-back localization well- a sound coming from directly behind the dummy head is difficult to distinguish from one coming from directly in front, and second, these recordings- and all others which rely heavily on time-of-arrival differences for localization must be listened to on headphones to work properly. If you listen on speakers, early reflections in the listening room can mess up the localization, and if there’s even a foot of difference in the distance from you to each speaker, the whole effect goes to Hell.
Incidentally, back in the days when I was still doing recording engineering for fun and profit I did a crude experiment on the role of time-of-arrival differences in placing sound within the stereo image. This involved using a mono in, stereo out digital delay to control the time difference between identical signals fed to the left and right channels of a pair of headphones. I found that if I tweaked the delay so that the left channel was a mere 100 microseconds ahead of the right, the source appeared to move from the center to between 10 and 11 o’clock. At 300 microseconds difference, the source appeared to be nearly hard left, even though the amplitude in each channel was equal.
Hm, I don’t have headphones, so I was forced to make due with just my computer speakers. Still, I was able to get some of the effect!
Whoops, that should be “make do,” right?
The primary method by which humans localize low frequency sound (under a few kilohertz) is by the arrival time. The intensity difference only becomes dominant in higher frequency sounds. 100 microseconds is actually a long time, it only takes a discrepancy of 10 microsecond or so to make a sound seem to shift to one side. If a sound is straight ahead we can tell if it moves just 1 degree of to either side (this increases to 10 degrees for sound at 90 degrees).
One interesting thing is what happens when you ignore the head. If you just add a time and/or level difference between the two ears what you end up with is a sound that is perceived as being inside the head. It may be shifted to one side or the other, but you cannot make the sound appear to be outside the head using only time and level differences. That is one of the ways things like the pinna, shape of the head, hair, and other similar cues come into play.
The issue of front/back confusion that Ktesibios brought up leads to another interesting problem. As he said, the front and the back cannot be differentiated based on just time and level differences. The time delay and level difference from having something come from 45 degrees to the side and front is exactly the same as something coming from 45 degrees to the side and back. However, something coming from straight out to the side but 45 degrees up will also be indistinguishable. Even more, something 45 degrees and 10 feet away will be the same as something 45 degrees and 30 feet away (assuming you do not know the “real” intensity of the sound). This is called the “cone of confusion”, for any given level or time difference there is a cone of locations that are impossible to distinguish. This makes determining the height of a sound particularly challenging. Once again the shape of the head, the ears, the hair, and such are primarily responsible for resolving such ambiguities.
@ Maronan (#23):
If you were centered between the two speakers you should get some of the effect. However, because listening on speakers introduces some crosstalk between channels (because each of your ears can hear both speakers), the pipeline between each of the dummy head’s microphone “eardrums” and each of your eardrums is somewhat polluted; consequently the full effect is not obtained.
Sitting off-center will have an even more pernicious effect. As we’ve seen, a time-of-arrival difference denominated in microseconds will have a strong effect on the perceived location of the source. Imagine that a particular source in a binaural recording was located so that the left channel leads the right by 100 microseconds. Now suppose that you’re sitting in a location such that the distance from your head to the right speaker is one foot shorter than the distance from your head to the left speaker.
At typical room temperature at sea level, sound propagates at approximately 1130 feet per second, so that one foot path length difference translates to a time-of-arrival difference of approximately 885 microseconds- so now the sound of the source mentioned above arrives at your left ear 785 microseconds after it arrives at your right ear. Where will you perceive it to be now?
We’ve managed to make amplitude-based stereo work fairly well- if it were not so, coincident micing techniques like X-Y, Blumlein and mid-sides wouldn’t work at all predictably- but they do. In fact, listening to a Blumlein pair on a set of good monitors in a decent control room can give very good perceived localization. In a world where listening setups vary haphazardly, simple solutions like amplitude stereo, crude and brute-force though it is, are more reliable than methods like binaural recording.
Wow. My headphones are labelled L and R, and I double-checked that I had them on correctly, but all the effects were on the wrong side.
Apparently my headphones are mislabelled. I should sue the manufacturer!
Very cool indeed.
Unfortunately the dang algorithm doesn’t seem to work as well underwater. Possibly because as primates we have evolved to detect the source of sound based on it’s speed in air? Any evolutionary biologists care to chime in on that subject?
I tend to have a harder time distinguishing the direction
of the source of a sound when I’m diving.
http://library.thinkquest.org/28170/36.html
“Relying on one’s hearing, it is extremely difficult to orientate oneself under water. In the air, sound reaches one of the ears .00003 seconds earlier than the other. This fact allows the source to be identified within an error of 3?. Because of the high speed of sound under water, it is perceived by both ears virtually simultaneously and the orientation error may reach up to 180?. Bad orientation under water is also due to the prevalent bone conductivity. Sufficient audial orientation is possible to be acquired only after systematic training. After training ceases, however, this ability disappears.”
Speed of sound in air
http://hypertextbook.com/facts/2000/CheukWong.shtml
Speed of sound in water
http://hypertextbook.com/facts/2000/NickyDu.shtml
I remember in a Japanese themepark they had a set up like this. you go into a windowless lightless room and put on headphones. Then a holophonic sketch stakes place – its usually something freaky. It was all in japanese, but I got that it was a guy going into a hospital only to find more than he bargained for, it was about 10 times more chilling than virtual dentist would be. The girl I was with almost peed her pants.
@ Ktesibios
There is a more serious problem than just crosstalk. The way this recording is to reproduce the effects the head would have on the signal. If you are wearing headphones, this is important because the sound goes directly into the ears and avoids the head. When you are using speakers, however, the sound interacts with your head in real life. So they applied filters to make the sound seem to come from one direction, but since the speakers in a completely different location you get a second set of filters applied. Further, you are get two sets of time delays and two sets of intensity differences layered on top of each other, one set built into the sound and the other from the locations and orientations of the speakers. This leads to a muddling of all the location cues. If your speakers are set up well it should compensate somewhat for the time and intensity issues, but it can’t fix the frequency problems.
“It’d be great to see this sort of effect integrated into games.”
Games have used techniques like this for years. AM3D make the best engine, but Creative’s more basic EAX is standard on most soundcards these days.
Re: #12 “I experienced something like this at Disney World yeeears ago. It might even have been exactly this, because i remember a (virtual) barber and the very distinct sound of scissors cutting hair. I wish i could remember which park it was…probably MGM, as I believe it was in an attraction about movie magic. You went into a sound booth, put on headphones, and listened to the recording. There was also an exhibit on Foley artists in the same attraction.”
That was “The Monster Sound Show” at MGM, which closed around 1999 (more info at http://www.waltdatedworld.bravepages.com/id69.htm )
The building currently has another binaural attraction “Sounds Dangerous” with Drew Carey. ( info at http://www.wdwinfo.com/wdwinfo/guides/mgm/st-sound.htm )
The MouseTunes podcast episode at http://www.mousetunes.com/podcasts/mousetunes_071606.mp3 has a binaural recording of it.
In the Magic Kingdom in Florida, there was a show called “ExtraTERRORestrial Alien Encounter” that has been changed to “Stitches Great Escape”. Both of these shows used binaural sound. You can hear binaural versions of both shows in the wdw-memories podcast episode at
http://wdw-memories.podomatic.com/enclosure/2006-10-29T05_27_02-08_00.mp3
Over at Disneyland, the most recent version of “Great Moments with Mr. Lincoln” had a binaural experience, complete with a haircut simulation like the one in this post. Mr. Lincoln is currently hidden behind a screen as his theatre in Anaheim is being used for a film about the 50th anniversary of Disneyland. VisionsFantastic has a binaural recording of it at http://www.visionsfantastic.com/main.php?page=lincoln but you have to create a free account to listen to it.
There is a weekly podcast that does binaural walkthrus of Disney attractions at both North American resorts at
http://windowtothemagic.com/
This method is typically employed in 1st person video games, (and has been for about 4 or 5 years now).
Most people assume that the games are all visual, but adding spacial effects to the audio portion of the game creates quite a sense of realism.
Yeah, video games do use it. I loved how it was used in Thief III, a first person sneaker game where you often had to rely on sound to guess where the guards were so you could avoid them.
More such 3D audios are @
http://funnylinkz.blogspot.com/search/label/Audios