Ralph Glasgal

I thought this was a group interested in computer/DSP audiophile techniques. Nothing is perfect. But with modern processors it is easily possible to come up with a reproduction method that sounds a lot more like live music than does the 60 degree loudspeaker triangle even when playing the existing library of LPs and CDs. If you also use more modern surround microphone techniques (See my paper on the design of the Ambiophone) then you can leave stereo in the dust as far as sonic realism is concerned. But color digital photography is not going to replace oil painting either if it is a matter of taste and they are deemed equivalent except for a few minor defects. Comparing a 90 year old technology which is subjective, rather than mathematical or based on psychoacoustic rules, to a modern physics based technology by simply stating that the differences are so minor as to just be matters of taste is, I guess, also matter of taste. The prejudice among audiophiles regarding DSP does not make much sense since every new recording or download they listen to is DSP processed.

Sorry, wrong example. I did forget about the pinna localization mechanism which does let you localize for high frequencies with just one ear facing a stage. So standing sideways you could still get some idea of where the instruments are on the stage. Yes the pinna function is the spectral part of the HRTF as opposed to the interaural head shadow part which is what stereo relies on entirely. But in the case of stereo speakers this sideways pinna sense does not work since there are only two pinna cues, one at 30 degrees to the front of you and one at 30 degrees toward the back of you. Not a great way to localize without any ITD or ILD. When you are sideways to stereo speakers the brain gets two fixed loud pinna patterns from one ear and another lower level pair from the shadowed ear. I doubt any two humans will hear the same thing under these circumstances but mostly you just localize tor the two speakers with nothing in between. The internalization effect of headphones/earphones is indeed due to the absence of localization cues that are not the interaural ones of level and delay. The only other significant part of the HRTF left then is the pinna functionality which is impaired by any type of earphone. You can test this yourself. Hiss out of your mouth and bring a finger into both of your ears close to the entrance to the ear canal and see what happens. Do it again with just one ear and let us know what you hear. If the speaker at the back of hall has no delay and is loud you will localize to it in most cases. If sideways the source will vary from left to right with slight motion if the levels are the same where you are standing.

Adamdea is very creative so I am not sure I follow all of the above. But in a concert hall if you stand up and turn sideways to the stage you still know where the stage is, but you have lost the ability to tell where the instruments are on that stage left to right. Likewise in a stereo system, or in a binaural loudspeaker system as you turn your head sideways to them you lose horizontal plane localization. With earphones this does not happen and this disturbs researchers more than listeners, I think. This earphone effect is sometimes blamed for internalization but that is wrong. (See the Washington University Course link above somewhere)

Thank you! This is a really fine reference with excellent graphics. Anyone interested in localization and how the ear works should read this.

it is a myth about the head in a vice. You have more freedom of motion or better location in the binaural loudspeaker systems like ambisonics, Wavefield Synthesis, Ambiophonics, etc. than in stereo. I rotate my head, lean, nod, stand up, move my seat back many feet with no vice in sight.

Really good question which was maybe answered only about 60 years ago. The pinnae act as direction finders over a full sphere of angles. So you can only detect height if the sound has a lot of frequencies in the 1500 Hz and up range. You can detect height using just one ear, try it using running water or a pet cricket. There are lots of pinna patterns in the literature that show which patterns the brain uses to detect height. They differ from one individual to another and that is why VR systems are having a tough time imitating this mechanism.

Good try. Stereo main mics have to be close to the direct sound to avoid a sewer effect which occurs in playback if all the hall ambience now comes from the front speakers mixed with the direct sound and having the same directional properties. The home room reflections have little to do with where the mics are placed.

The problem with earphones is that using in the ear canal phones mean the pinna are not functioning. This has the effect of making the stage seem to be inside the head, extending from ear to ear. Circum aural phones also similarly interfere with the pinna and all phones make all the upper frequencies come from one direction, likely 90 degrees. With head rotation the entire front stage moves with you. Part of the HRTF is the pinna direction finding function, so indeed you can use a computer/DSP device like the Smyth Realizer to restore the pinna function and provide externalization. You need to be technical and rich to use the Smyth, however. One eared individuals can indeed localize if the sound has high frequency energy that can be used by the pinna direction finding mechanism. This single and dual pinna ability is documented by hundreds of papers over the last 50 years or so. See my .org website for a lot on this. In a normal binaural soundfield, if you rotate your head, sound sources do not change position. This is not true for earphones, but is true for loudspeaker binaural systems of which there are many available now. Note 2.0 LPs and CDs are largely binaural in nature since the mics used do not know they will later be reproduced using Antique means. For a summary of how to address all these issues you can read the tutorials and the book at www.ambiophonics.org. I try to collect all the papers on these issues here with links, etc. as a charity.

The word binaural has somehow been preempted in recent years by the earphone world. Actually normal every day, two eared hearing is termed binaural as opposed to monaural where only one ear is exposed to sound. Blumlein called stereo binaural in his original patent. it is indeed possible to have loudspeaker binaural. Examples include Ambisonics, Wavefield Synthesis, Ambiophonics, BACCH, and more appearing now every month. Basically, if what reaches the ear by any means has the same wavefield as one hears every day then it is a normal binaural soundfield. I would add that earphone/headphone binaural is not truly normal binaural since the phones interfere with the soundfield hitting the pinna. So like stereo via loudspeakers, binaural via phones is not realistic binaural.

I should probably not get into this, but there are some real technical psychoacoustic issues here, not just myths, which STC has so ably hinted at. Remember however that stereo is also an art form and thus does not have to be realistic in the concert hall sense. So here I will discuss only the realism issues. In a concert hall if a trumpet is at stage left you will hear a time difference at the ears of about 700 microseconds and a level difference of maybe 7 dB So let us assume a perfect mic and recording media have captured these values exactly including perfect frequency response, perfect resolution, and exact sound level. Now we play this back without error through perfect amplifiers and two perfect speakers set 60 degrees apart. Well, apart from the fact that there is no proof in physics or psychoacoustics that this angle is correct, we can easily prove that it has a lot of real defects and distortions. First the 700 microsecond time difference recorded is reduced to about 220 microseconds so that side trumpet is now in the middle of the violins. The original level difference is likewise reduced at the ears by half since both ears easily hear both speakers. No matter where a sound was originally located, the pinna see only the pattern produced by sound sources at 30 degrees, not the almost 90 that the trumpet produced in the hall. I could go on about all reverb now being frontal, the peaks and dips in the frequency response, central bass doubling, but enough. So these realism issues cannot be corrected by any mic technique since the problem is mostly one of reproduction and not predictable since listening angles vary so much as does head size. However, if you get into the subjective art form realm, then you can tweak recordings to make up for one 60 degree defect or another. But it is tough to get two human beings to agree on which tweak sounds more realistic or purer or whatever. See www.ambiophonics.org for endless papers and tutorials on this subject and how to fix the problem if you want binaural (normal hearing) realism.

Good question. The impulse responses are just lists of reflections from particular walls in the hall. The ones that are most important to the brain are those from the sides, the rear and the ceiling in that order. Ceiling reverb is mostly mono so the brain cannot do much with it. So you take a left channel front signal and convolve (process) it with an impulse response from say 120 degrees on the left side and send that signal to a speaker at that location. and repeat for as many IRs and speakers that make a difference to your ear. I use 24 at the moment but that is really a waste of money. So there are some speakers at the sides but they are not emitting any direct sound just reverb that an IR says should come from there. You can do this for any recording including 1.0, 2.0, 4.0. 5.1 etc. Putting a mono LP or 78 in a hall is a revelation for solo voices or piano or violin sonatas or old jazz combos. If you put direct sound at the rear willy nilly, you can get comb filtering, front to back inversions, and no logical change in the sense of space. But you can put 2.0 direct sound into two rear speakers at a narrow angle behind the listening area using a miniambio or computer app to get a sense of envelopment, great front stage width and depth, but not a sense of being in a hall. You can read the Envelophonics tutorial to see how easy it is to do this if you are willing to use two small speakers in the rear. Ralph

My pleasure and this is a good question. Any side speakers are those producing hall ambience from hall IRs so they never have any direct sound. If the rear speakers are using the Envelophonics protocol then they do reproduce direct frontal sound but you cannot hear this, only a sense of envelopment. Otherwise the same or a different rear pair is producing the rear pair of a 4.0/5.1 recording. In 2.0 file reproduction the brain does need to have a rational field of early reflections coming from directions that are not frontal and these very early reflections, like those from heads, and seatbacks, need to have directional cues intact, thus you need XTC for this rear pair. I know, this is not easy to get used to.

This is not an easy technology to grasp and there are a lot of options so I am not sure I understand the issues above. For existing 2.0 media like LPs or CDs you can place such recordings in a domestic concert hall by using 3D impulse responses to drive a large number of surround speakers. Nothing is added to the front speakers. However if you crosstalk cancel the front pair you will get a much wider/deeper stage and much flatter frequency response to go with the surround speakers. There is no way to add ambience to the front channels in recording or at home without the risk of getting the sewer effect. That is why 2.0 recordings ate made with the stereo mic at the conductor or close to the players, well within what is called the critical radius where the direct sound equals the reverberant sound. Also the brain will not accept frontal hall ambience as real. For any frontal pair in LPs, CDs, and 4.0/5.1 media you can enhance frontal stage envelopment by a process I call Envelophonics using two speakers conveniently behind the listening area. These speakers emit a crosstalk cancelled flat version of the frontal pair and eliminate a static listening room reflection problem and a pinna localization cue bias. Tutorials on all this as will as free 3D IRs at ambiophonics.org. So there is some direct sound at these optional rear speakers but the effect is a wider, deeper, more natural front stage. e-mail me at [email protected] if you wish. So the ideal for reproducing 2.0 media is to use a frontal and an enveloping rear pair for direct sound and a set of 3D hall IRs for side/rear surround ambience. The ideal method for reproducing 4.0 media like SACDs, 5.1/Auro/Atmos/DTS/etc. is as for 2.0 but adding one more XTC pair for the rear direct sound stage. If the rear pair of a concert SACD is hall ambience, then the third rear XTC pair will provide a full half circle of rear hall sound and you don't need to have additional surround speakers. In the case of movies, sometimes it does sound more realistic if the hall surrounds are operating. If you are into height reverb then you can use these speakers with both music and movies.

But it is not just mere localization that is at stake. In a real hall, all the localization cues are consistent. In a stereo system they contradict each other. The result is a sense that the soundfield lacks realism. Also it is not flat in frequency response in the range where the pinna are active even if the response leaving the speaker is perfect. It is like printing a perfect color negative in black and white and expecting the brain to see color..

I should have said up to 170 degrees. It is easy to adjust the angle of the orchestra if you don't like 5th row center. If you add the surround speakers for hall ambience you can sit in the balcony. Stereo limits you to 60 degrees max with no control. I once had a patent on a stereo width control but it only went from 60 to 0 degrees. I think though when you see the kind of detail, clarity and depth you can get when you reproduce all that is on an LP or CD without localization cue distortion that you might want to move up a few rows.