Matt and Don discuss Spatial Audio with Edgar Choueiri, Pt II
HTML-код
- Опубликовано: 13 окт 2024
- Matt and Don will delve deeper into how we perceive sound in acoustic space with one of the leading experts in spatial audio reproduction and realism. Dr. Choueiri previously discussed the fundamental principles of spatial reproduction and the shortcomings of many existing systems. In this video we take the topic deeper and discuss some of the current cutting edge methods of spatial sound reproduction.
Watch How to Achieve Realism in Music Reproduction: • How to Achieve Realism...
Shop Audio Advice: bit.ly/3thUFI1
Recommended Subwoofers:
SVS SB-1000 Pro: audioadvice.io...
SVS PB-1000 Pro: audioadvice.io...
SVS SB-3000: bit.ly/SVSSB30...
Paradigm Defiance X12: bit.ly/Paradig...
Recommended AV Receivers:
Denon AVR-X3700H 9.2CH AV Receiver: audioadvice.io...
Yamaha RX-A2A: audioadvice.io...
Denon AVR-X6700H 11.2CH AV Receiver: audioadvice.io...
Anthem MRX1140: bit.ly/MRX1140
Monolith M8250x 8CH Amp: bit.ly/8CHamp
Recommended Speakers:
SVS Prime 5.1 Speaker / Sub System: audioadvice.io...
Polk R700 Tower Speakers: audioadvice.io...
Heco Aurora 700: bit.ly/HecoAur...
Heco Aurora 1000 Tower Speakers: bit.ly/Aurora1000
Klipsch RP-8000F Tower Speakers: audioadvice.io...
Arendal 1723S Tower Speakers: bit.ly/Arendal...
Focal Kanta No2: bit.ly/Focalka...
Sony SSCS5 Speakers:audioadvice.io...
SVS Prime Elevation Surround: audioadvice.io...
Audioholics Recommended Cables:
250ft CL2 12AWG Speaker Cable: amzn.to/2vwS9QH
50ft 10AWG Speaker Cable: amzn.to/3DeMTT5
Locking Banana Plugs: amzn.to/2ZQt15x
9ft 4K HDR HDMI Cables: amzn.to/2WiIXeD
Follow us on:
Patreon:
/ audioholics
FACEBOOK
/ audioholics
TWITTER
/ audioholicslive
Audioholics will never cease to keep my interest! Keep up the amazing work Don and Matt!
I actually have the BACCH4Mac DSP. I have been an audiophile for nearly 40 years. This is the single most substantial and revelatory upgrade I have ever experienced. This is a total game changer.
Really amazing video guys keep it up
I love this! Please keep this kind of content coming!
This is incredibly interesting.
Here's what I think is the history of phase cancellation in audio. (I'm an inventor of a competing technology, more about that later.) My first encounter with phase cancellation was with the Zenith method for compatible FM stereo multiplexing which the FCC adopted as the standard. FM broadcasts were monophonic L+R up to 15 khz. An L-R signal was broadcast at a higher frequency and after the FM detector the L-R signal was separated. In the "matrix" decoder L+R and L-R were added in phase and out of phase to get 2L and 2R.
Somewhat later Ralph Glasgal working for Fisher Radio invented and patented a "stereo separation control. This control was used to feed the monophonic signal out of phase to varying degrees to the L and R signals reducing whatever they had in common. This is probably the same thing you see in HT receivers that have a "wide" function. The increased separation is desirable to enable a monophonic center channel to the front speakers for a broader presentation and to keep the dialog reasonably close to the viewing screen. I call HT son of Quadraphonic sound, the only real difference besides the format which allows for total channel separation of all of the 4 channels is the addition of a center channel and an outboard subwoofer or two. (It should be noted that even in purely electronic circuits where negative feedback is correctly used there are limits. When the feedback signal reaches a frequency where it is 180 degrees out of phase with the input signal and the loop gain is one or greater you have an oscillator.)
Once the concept is extrapolated from electrical signals to acoustic fields you enter a whole new realm of complexity. We are now into the realm of time varying three dimensional vector fields where geometry becomes a factor. As far as I can tell Ralph Glasgall was the first or among the first to experiment with this but not before he used a wall between the two speakers as Edgar Choueiri suggested. Since that was impractical he advanced to cross channel phase cancellation. Because of the relatively slow velocity of propagation of sound and the difference of distance between each ear and the two speakers a time delay has to be introduced into the cross channel out of phase signal and of course it must be of the precise loudness so that it will arrive at each ear at precisely the right time and loudness to effect the maximum cancellation. This happens only at a very limited point in space and room reflections will compromise the effect. The problem here is that as frequency increases the wavelengths get shorter until you reach a point where the area of cancellation is so small that for all practical intents and purposes it no longer works. As the phase differences are no longer in sync the cancellation effect diminishes and once you reach 90 degrees the signal not only doesn't cancel it adds back in. This is why the uncorrected process will have a high frequency peak. So to tame this peak a filter or equalizer is required to flatten the frequency response.
I heard Ralph Glasgal's system he calls "Ambiophonic" (not to be confused with Ambisonic which is an entirely different technology) on two occasions. The experience was interesting but it was not a solution to the problem of high fidelity sound reproduction IMO. What I heard was very precisely located sound sources about 150 degrees wide on a line between myself and the speakers. He also had a series of other speakers surrounding the listener that was supposed to create the reverberant field of a concert hall. It didn't work at all IMO. It was obvious that Mr. Glasgal had spent a fortune on this system and if you told me it cost over one million dollars based on the equipment I saw, I'd believe it. One question I asked Ralph he couldn't or wouldn't answer to may satisfaction was why his time delay circuits had a recursive topology. All he'd say was that it worked better that way. More about other technologies including my own and discoveries I made in posts replying to this one.
One of the first flaws I noticed about quadraphonic sound was that I could hear where the so called ambient sound was coming from, the rear or side speakers. That is exactly what you don't want at a live concert and if you do hear where reflections are coming from it is an unacceptable acoustic effect. Classical acoustic science defines a parameter called diffusivity index, more about that later.
So to design a system that will not create this effect you have to understand how the human brain senses directivity. The explanation given early on here considers three factors. The first was ITD, interaural time delay. This is the time difference between when a sound arrives at one ear and the other. The second was ILD, interaural loudness difference. This is the difference of loudness of a sound between your left and right ears. The third one added later on was HRTF head related transfer function. This is the difference in spectral balance created by sound diffracting around your head. The explanation is dead wrong. The proof is that binaural recordings made with microphones in a dummy's ears or in your ears if you are the dummy and played through headphones meet all three criteria yet it fails as was pointed out here. When you turn your head the sound turns with it unlike normal sound. So close and yet so far. Question, why does that matter? The answer gives enormous insight into sound and hearing. The reason is that the sounds you normally hear are vectors which have a direction of arrival, the sound from headphones are the equivalent of scalars which always arrive the same way no matter how your head turns. So how does it work? The same way two dimensional moving objects seem to be in three dimensions using Disney's multiplanar camera. Your brain is wired for short term acoustic memory. When you turn your head the sound from one direction arrives slightly sooner in one ear and slightly later in the other ear. Depending on which way you turn your head and which ear hears it sooner or later determines not only the direction of arrival but whether the sound is in front of you or behind you. So with head tracking and both signals delayed in time when the camera senses that you turned your head one signal is delayed slightly more and the other slightly less. This simulates how headphones can accomplish this by emulating vectors but usually only in one plane. But it has to be fast. The time delay difference that is detectable is from two to five microseconds. I discovered this nearly 50 years ago. So now the criteria for the vector reverberant sound field has one parameter that must be met, diffusivity index DI or the reverberant field won't be convincing, it will have the same problem quadraphonic sound had.
It should be obvious that you cannot emulate the sound of a loudspeaker with headphones. I am really surprised at the suggestion that you could. Even if a speaker has no level controls that adjust the relative loudness of the drivers or their frequency contour the same speaker can sound a seemingly infinite number of ways depending on many variables. Some of them are the acoustics of the room they are in, the location of the speakers in that room, the direction they are pointed in, the location of the listener, the electronics it is connected to, even the type and length of wire used to connect them to amplifiers. So for the same speaker there are a seemingly infinite number of ways that speaker can sound. So how can you possibly claim to emulate the sound of any loudspeaker when it doesn't have a specific sound ever? The on axis sound it would produce in an anechoic chamber only eliminates the variables of the room and BTW that sound while useful for making measurements is horrible for listening to music. Sorry Edgar, no cigar for you on that one. BTW the tactile impact of deep bass from a really great low frequency reproducer like my own Teledyne AR 9s can never be produced by headphones.
With respect to the dots; in a great concert hall, opera house, cathedral many hundreds, even thousands of people enjoy sound far better than any home hi fi can produce. Even if this idea works, it can only work for one person at a time. It has a sweet spot that moves with just one person. Is that even acceptable? It means you can't share the experience with anyone else. How sad that listening to music as well as this can do (I'm not saying yea or nay, just pointing this limitation out) is a solo experience. The reason is that this system does NOT recreate the original sound field but only one that might seem like it in one spot at a time. We don't even know if the original space where it was recorded is optimal for the type of music on the recording. And of course the existing vast library of recordings is not compatible with the process because the microphones were not located in a place that met the criteria of this method.
The problem with any system that tries to create a reverberant sound field whether it's quadraphonic, 5.1 which I call son of quadraphonic, X.Y which is grandson of quadraphonic is that as soon as you point a speaker at the listener you'd better have an awful lot of them because each time you move half the distance closer to any one of them their loudness increases by 6db. As I pointed out above in a reverberant sound field in a space suitable for listening to music or even speech the diffusivity index DI must be very high. If you can hear the source of what is supposed to be a reflection it is ZERO!.
I call this the supermarket effect. In a large supermarket there can be hundreds of speakers in the ceiling. These are used for background music and for messaging such as "code five, spill in aisle 17." Wherever you are in the supermarket the sound will appear to come from the direction of the speaker closest to you on the ceiling. This is exactly what you don't want to happen for concert hall reverberation. For the reverberant sound field to be convincing the sound from any one direction must be overwhelmed by the sound coming from all other directions. If you were in a concert hall and you put your ear up to a wall when a symphony orchestra is playing its loudest fff you'd hear NOTHING. Yet in aggregate the sound that is reflected from the surfaces constitutes 90 percent or more of all of the energy that reaches you no matter where you are in the audience. If you don't believe a single other word Dr. Bose ever said, you can take this measurement he made at Boston Symphony Hall to the bank. The fact has been demonstrated countless times by acousticians countless times.
This flaw was the first thing I noticed about quadraphonic sound and it was instantly a deal breaker. Learning how to mathematically model, analyze, engineer and even measure sound fields accurately are no easy tricks. It's hinted at in my US patent 4,332,979. Interestingly the drawings of the elements for making measurements were inadvertently left in. They should have been omitted because the patent office said it was a different invention. You can see the similarity between the spherical array of unidirectional microphones shown and those others have built. Unfortunately one such paper with a video of results using this idea but with a different method, an impulse signal superimposed on a half dome video of a concert hall showing the reflections in the first 1/3 second slowed down by a factor of 200 is no longer on AKU's web site. The name of the paper was "Visualizing Reverberation." It is part of what I saw in March 1974. For technical reasons the impulse response is insufficient to measure both temporal and spectral aspects of individual reflections in a way that they can be segregated to each arriving reflection. This is critical because as sound decays in a concert hall its spectrum changes. Typically the RT at 1 khz is in the range of 2 seconds but by 8 khz, the limit of tests it's closer to 1 second. This is critical to understanding the tonality of music heard live and why if you don't include this in your method you will never get the tone close to live.
If 5.1 is the son of quadraphonic sound and X.Y is its grandson then Ambisonics sound is its great grandson. It was an attempt to record and play music using some vector algebra originally limited to four channels probably due to the lack of available recording channels for common formats of the day for practical use like most prosumer magnetic tape recorders. It had three pairs of cardioid microphones in each of three axes, one at each end. There was Left minus Right, Front minus rear, and top minus bottom. A fourth omnidirectional microphone was used to obtain the proper listening loudness. Later when more recording channels became feasible additional pairs were added. Ultimately today an unlimited number of recording channels, one for each direction of arrival from a spherical array of directional microphones and a corresponding playback channel for each one is possible. You'd need an amplifier and speaker for each direction. en.wikipedia.org/wiki/Ambisonics
Wave Field Synthesis is an entirely different matter. Here the sound is recorded in an anechoic chamber and the sound field is reconstructed using an algorithm derived from WFS is based on the Huygens-Fresnel principle, using the the Kirchhoff-Helmholtz integral. en.wikipedia.org/wiki/Wave_field_synthesis I've studied this carefully to determine whether or not it is identical to my own method I call Electronic Environmental Acoustic Simulation derived from vector calculus, Fourier transform theory and mapping functions, and acoustic energy field transfer functions. (Sound is a time varying pressure gradient vector field.) It's close but it has one problem. Notice that the solution is a double integral. Mine is a triple integral. The problem with WFS is that it has no way to segregate each reflection insofar as its time of arrival and its spectral change from the first arriving sound the EEAS does. Therefore EEAS can precisely describe every last reflection arriving at a point with complete precision. Expanding the vectors backwards in time to a larger closed surface it is essentially identical to WFS.
Adapting EEAS to play commercial recordings required some engineering skills including paring it down to its simplest form as it was originally conceived of as a very complex laboratory. I had very little hope for much in the way of results giving what I knew I was up against. So how does it work? The main two front channels are the same as a conventional stereo system. A real time computer, in the case of the current prototype a hybrid analog/digital computer uses the algorithm to generate the signals fed to an array of small speakers around the perimeter of the room firing their sound upward using the walls and ceiling as reflectors and diffusers yielding a very high DI and a very high listener envelopment. The reverberant field is a true vector field whose parameters can be adjusted in many ways. The system is tuned to the acoustics of the room once and then again for each recording. The settings are made by memory of live listening experience. Therefore the method is both objective and subjective since there is no one right answer. We don't know what the acoustics were of the original recording sounded like live and it may not have been optimal. Therefore the adjustments are entirely arbitrary.
Ultimately when it comes to room acoustics you have three choices. You can ignore it at your own peril and trust to luck, you can fight it and IMO it is a hopeless battle you are going to lose, or you can exploit it by incorporating it into your design by making it part of the system. I took the last option.
i coincidentaly achived binaural loudspeaker audio in my room:
i have a set of desktop speakers and rear speakers for 5.1 surround.
when i run a stereo signal through the front, and the flipped but same one through the back, so that the l-l/r-r lines have a crossing point in the seating position, my small room creates room nodes that somehow end up canceling crosstalk where i sit.
so its like headphones, while sitting with speakers.
on my system, i have to reduce stereo seperation to 1/3 to have a properly scaled binaural experience though, as my desktop speakers are about three times as far appart as headphone drivers would be. the sweetspot for proper binaural is tiny, though the acceptable sound field is decent.
when i move to a side wall, so that the front and back sets are now at the left and right side of the room, the nodes make it sound like i have a stereo sound source on the empty wall infront of me. actualy dissapearing speakers. the sweetspot is decently large, with the acceptable listening positions covering the entire wall.
the binaural experience sounds like its happening in my room/ a bigger space with my rooms acoustics though. speakers, compared to headphones, are not room agnostic. the binaural illusion persits though.
the "headshaving " binaural test sounded scarily real. the room sound in fact added to its naturality.
with non binaural stereo recordings i started to use this effect instead of stereo to 5.1 upscaling. it feels like sitting in between the musicians, as the stereo image plain is pulled to my sitting position. for that i also keep the original stereo seperation, as head scale accuracy is less necessary and thus a larger sweetspot is possible.
in the sweetspot, there is also bass resonance subharmonics that get so low i can feel my eyes unfocus with strong bass notes, as my body starts to vibrate. visible bass impact so to say. i have to keep that relatively low volume though, as otherwise the central heating in my room starts to resonate audibly and that can't be healthy for the pipes.
I wish you would test some of the newer Pioneer Elite receivers with Dirac. I'm reading some great reviews on them. No test bench stuff though
Haven’t watched yet, is there not a part 1? Is this perhaps miss labeled? I don’t see a part 1
Is there an advantage of combining surround sound for ambience (Atmos, Auro-3d, etc) with binaural together (for distance perception and specific location of objects in 3d space) versus binaural over speakers with XTC alone?
I've combined atmos 9 base layer sony avr with a denon 6 heights of auro. The sony has 4 heights in atmos as well. The denon has up to 4 subs and the sony 2 subs. It's very powerful for me. I'm using the auro3d center height, vog top center along with the front and rear heights. The sony heights I place in optimal spaces and can use the 360ssm to blend it nicely. I also use on the sony the center raise feature that brings up the center channel to the front heights, it's a very nice dsp sony uses for that.
It seems like absorbing the right side first reflection from the left speaker and left side first reflection from the right speaker would add to better imaging and soundstage width in typical stereo playback.
Not likely. In a normal stereo system the goal is fundamentally different from what we discussed here. With normal stereo we are trying to recreate the soundscape and all the spatial cues within the room. Lateral reflections are an important part of perfecting that space.
By contrast, this approach is attempting to eliminate the room and beam the ear signals directly to the ears.
Matt, thank you for the response. My idea was that a second best option to keeping a physical barrier between 2 speakers, not realistic in normal living conditions, would be to try to absorb as much energy coming from the first reflection into the "wrong" ear. Is the cross talk cancellation only viable when done to the first, fastest arriving and loudest, wave? I understand that it's the first signal that gives our ears spatial cues.
@@williamkramer9069 it’s not reflections that cause crosstalk per say. Crosstalk exists in an anechoic chamber. Crosstalk simply refers to the idea that sound emanating from a given speaker is heard by both ears. In the cause of binaural, that isn’t what we want. We want the left ear to hear only the left speaker. Not the right speaker. Wall reflections are not the cause but can contribute.
when the headtracker requires a wearable, going straight for the much simpler headphones makes sense.
headphones can be combined with external subwoofers for the body experience.
headphones can also be used to support low volume external soundsystem playback.
The head tracking does not require a wearable. Head tracking is done very precisely with a simple webcam.
@@PoesAcoustics that version has merrit
In your opinion what is the best high end headphones for gaming and music I’m leaning towards Audeze lcd gx what are your thoughts,
What happened to Gene? Did he take the day off.
Gene doesn’t join the Matt and Don show.
@@PoesAcoustics ok
I was lifting weights while they were geeking out. 😆
@@Audioholics good to see your maintaining your health.
promosm