[bpape]

As you know, the head itself casts a shadow, so the radiation pattern will be rather cardioid, with differences from front to back ranging from <5 dB to as much as 12 dB depending on the frequency.

Exactly.  And that's part of why reflections can be so damaging.  The reflected sound is not of the same response distribution as the original sound so it not only smears things in time but also mucks up (technical term) the frequency balance.

Bryan

[youngho]

So...I think we agree that the observation "Real sounds tend to disperse in all directions" should not be used to argue for omnidirectional speakers and Bose 901s uber alles, not that Darren Yeats was arguing that.

[darrenyeats]

As you know, the head itself casts a shadow, so the radiation pattern will be rather cardioid, with differences from front to back ranging from <5 dB to as much as 12 dB depending on the frequency.

A classic box loudspeaker is designed with the aim of minimising cabinet resonances and diffraction. In fact, box loudspeaker designers make a song and dance about how 'dead' their cabinet design is and so on. My mouth isn't particularly rigid or stuffed with dampening material (although my wife might wish it sometimes). Are the directional characteristics of a box speaker at various frequencies just like a person speaking? I question that.

Of course the lower the frequency the more the tendency of a sound to spread in all directions, whether natural or reproduced.

Now the higher the frequency of a sound, the more stark the character of its directionality. When I think of the highest frequency sounds in music I think cymbals first (and triangle second). This sound spreads in all directions. This instrument is (a) high frequency therefore directionality is very important in reproduction and (b) disperses in all directions in real life. This doesn't strengthen the argument for box speakers - cymbals are kind of important.

Exactly.  And that's part of why reflections can be so damaging.  The reflected sound is not of the same response distribution as the original sound so it not only smears things in time but also mucks up (technical term) the frequency balance.

I used to see it that way, but not anymore. You can view the sound that arrives at your ears two ways.

First way is to view it as it would be recorded by a microphone. Everthing about this waveform will be mucked up by reflections, frequency response being only one aspect. (DRC is aimed at correcting these things, but can't, because you can't correct reflections fully by altering the output of the speakers.)

The second way of viewing it is not how it looks in a scope, but how it is perceived by a human listener. Now human listeners have a lot of in-built 'DSP' for dealing with all kinds of crap that happens with real sounds and real spaces. We are filtering reflections all the time in real life. When someone talks in a normally reflective room, do we perceive several other people talking the same words but a bit quieter? Do we perceive the person speaking as having a strange voice? No. In fact, when we are truly listening to someone's words or focus on their voice in a normal acoustic, we don't hear reflections at all although they are surely there! Only in an atypical acoustic do we suddenly become aware of reflections or their absence. However, if the person were to sit in the extreme corner of a normal room, the corner might act as a horn and cause us to perceive a change in quality of the voice. Voices are perceived to change likewise, e.g. frequency balance, out of a speaking trumpet. This is because reflections occur too near to the source and are perceived as the source itself. If the source is far enough away from a reflective surface, magically the originating sound becomes free of the reflections and we hear it naturally (including e.g. its unadorned frequency balance). Is this is a property of the physics of sound waves? No way! Is it a property of human sound perception? Yes. Of course, we hear the reflections too (which contribute to the frequency balance recorded by a microphone) but since these are sub-consciously filtered out when we are truly listening to something, they are not perceived consciously.

Reflections can be heard, but only with putting forth conscious effort.

Such perceptions are possible only because we have a lot of sub-conscious processing going on between the ears and the conscious mind.

Now, consider the problem that "reflections happen" anyway. It is very difficult to stop reflections entirely, and most treatments do not work broadband. So whatever you do, you're left with reflections anyway! The question is, are these totally inevitable reflections/reverberations of a nature that the brain is used to filtering out?

It is better to have a normal amount of normal reflectivity, than to have a reduced but abnormal reflectivity. Several posters have commented on rooms that can be "too dead". Clearly the amount of reflection isn't so important as at first glance...rather the quality of reflection is more important.

If instead we create a reflective environment of familiar quality, the brain is much better able to recognise reflections for what they are and remove them from the perceived listening experience.

In other words, I am discussing a different way of thinking about reflections. I didn't invent this way of thinking, I merely present it, and state that for me it makes sense and I have got good results with it. Of course, if you don't agree with that way of thinking  - the key bit is the human, perceptual side of the argument - you won't agree with the conclusions either.
Darren

[bpape]

Agreed on some points.  The QUALITY of the reflection is key.  When someone talks, it isn't different out the side than it is from the front.  When a speaker 'speaks', it IS a different response - therefore, it's a bad reflection IMO. 

IF the source is far enough away from a boundary, I'd agree to a point.  BUT, in most residential spaces where we listen to reproduced music, we're not far enough away to make that work.  We still have a smearing in time and a change in the blended frequency reponse of the 2 waves.  Having an omni-pole simply cast so many reflections in all directions that one can't possibly distinguish them thereby APPARENTLY masking the problem isn't the answer IMO.  It's a neat 'effect' but that's all it is.  I think Bose tried that same idea with the 901?    Not a true omni but the same general principle.

As for the cymbals and triangle thing, that has nothing to do with it being high frequency or not.  It has to do with the fact that those two instruments DO radiate sound in all directions equally at all frequencies.  Something like a trumpet or clarinet do not. 

Did you ever wonder why panel speakers have that magic on strings that almost no box speakers do?  Think about it...  Because they vibrate and reproduce sound in exactly the same manner as the original instrument.

But at the same time, other instruments do not make sound in the same fashion.  Most of the non-stringed instruments simply amplify something else that's vibrating (lips, reeds, etc.) and spit it out the end of the instrument after having the length of the pipe changed.  Drums are different yet.  They are not omnipolar.  Some are dipole, some are monopole.

So, how do we get something to reproduce all of these different things accurately?  We try to reproduce what the mic 'hears' and how it hears it and try to eliminate the room from interacting with that reproduction.  We don't try to reproduce what people can't ignore and filter out.

Bryan

[darrenyeats]

I'm really appreciating this discussion. This is the sharp end of audio for me.

Anyway, on the drums / LF thing, I don't think it matters so much what kind of dispersion a drum has at source, because LF ends up coupling with the whole room. The room couples with the LF instrument, you get standing waves etc and perceiving direction becomes more difficult anyway. So drums might be dipoles or monopoles but it is more difficult to hear the difference. The differences are probably manifested by which room modes get excited, rather than a conscious perception of reflection.

My point with cymbals is that they sound predominantly in a part of the frequency spectrum where listeners are most sensitive to directionality. So, if directionality is a consideration with voices or some other instruments you mentioned, it must be even more so with cymbals.
Darren

[youngho]

Darren, there may be a few more aspects to this discussion that you haven't addressed yet.

Although I rather like Sonic Flare's "Sonic Circle" approach to defining listeners' preferences, I'm a simple guy, and the two basic questions for listeners, at least to me, seem to be "Does it sound good to me versus does it make me believe that it's accurate to the source?" and "Does it make me feel like I'm 'there' in the recording venue versus does it make me feel like the musicians are in my room?" Even if one feels as though one were in the recording venue, then it can break down to "front of hall," "middle of hall," and "back of hall." I think those who strive to believe that the reproduced sound is as accurate as possible to the recording itself generally prefer to hear the acoustics of the recorded venue over the acoustics of their listening room. In terms of loudspeakers, I think the designer who represents the epitome of this approach would probably have been John Dunlavy. Although Dunlavy tried to control the directivity of the different drivers, many other speakers with similar low-order crossovers have uneven directivity, thus requiring more care in positioning, and I think that it's these sorts of speakers that may benefit the most from the first-reflection absorption approach, and perhaps even more absorption still. This will probably have the effect of increasing the size of the "sweet spot," too. On the other hand, speakers with constant directivity tend to sound better in more positions in more rooms, as per Floyd Toole's research, because the quality of the reflections, if not absorbed, is more similar to the direct sound itself. However, as these speakers are placed closer to room boundaries, some absorption will probably be needed, though this can be bad if it's too thin.

Now, I never argued that a controlled directivity loudspeaker necessarily has the same radiation pattern as a human singer. The rather cardioid nature of the radiation pattern of the human voice has to do with the sizes of wavelengths reproduced relative to the size of the shadow cast by the head itself (and, in the vertical plane, the body). Even if a cymbal by itself is an omnidirectional source, the presence of a drummer to actually hit the cymbal in the first place will cast a sound shadow. However, if you aim to feel as though the musicians are in your room, and you listen to a lot of cymbals, then by all means, omnidirectional speakers are the way to go. In a sufficiently large room, with two-channel audio, where the speakers can be placed at least 2-3 feet (probably would be preferable for more distance for at least one wall, depending on the radiation pattern) from the nearest surfaces, and reflections are similar in nature to the direct signal, and the listener is sufficiently far from the back wall (or else significant absorption should be present on the back wall), then omnipolar and dipolar speakers may be nearly magical. However, there were a few prerequisites there that may not work for some listeners, and the majority of listeners don't have omnipolar or dipolar systems. Also, many people enjoy multichannel systems now.

I'm not saying that we should have different loudspeakers for different instruments. I wanted to point out that different instruments and voices have different directional effects, so one can't go from "real sounds tend to disperse in all directions" to choosing a particular loudspeaker style. With respect to reflections and absorption, as I wrote above, "It's definitely possible to have too much absorption, particularly if certain parts of the mid- and high-frequency audio spectrum are absorbed more than others. Basically, bass trapping is generally beneficial, and first reflections usually should be treated, but further absorption is not typically necessary" and "Normal rooms typically contain things like furniture, rugs, and bookshelves. Siegfried Linkwitz and Floyd Toole argue that these sorts of things may be sufficient for typical listening rooms. Others disagree." And listener preference trumps the above.

It's an interesting discussion, but because of the subjective nature of the experience, it's a little bit like arguing about wine. For me, I always keep things in perspective by remembering that my two most profound listening experiences were hearing the BSO play the Enigma Variations in Symphony Hall and listening to a recording of Nathan Milstein play Bach's Chaconne on a Sony clock radio.

[darrenyeats]

Youngho,
Thanks for your thoughtful and diplomatically written post.

For the sake of clarity, I am continuing to assume the speakers are around 1m or more from room boundaries (this is not an exact figure in my mind just one for the sake of argument). I think where we see things differently is over the issue of whether listening room reflections must mean the performance is heard as if in the room.

At first glance this idea seems obviously true. But then a question springs to mind.

Will the acoustic reflections in the performance space, when recorded from the microphone position and then reproduced from two central points in the listening space, be processed in exactly the same way as reflections at the real boundaries of the listening room?

The answer is not cut and dried in IMO. If you had a microphone at every point of the performance space boundaries and a loudspeaker at every point of the listening room boundaries in order to literally re-create the performance space reflections at the actual locations and time relative to the original instrument locations then the answer would be cut and dried.

However, that is not what microphones and stereo speakers do. Instead they record the effect of reflections at the microphone position in the performance space, then re-broadcast the effect from just two points in the listening space. The perception of the performance space is an illusion, not a literal reconstruction. This makes it possible to raise the question of whether these two kinds of reflections are really perceived the same way.

The above are questions. And I realise these questions prove nothing.

I merely say these questions are why I think there may be more to it. It could be that the brain filters listening room reflections differently to recorded reflections. This might enable a listener to "hear through" a listening space and perceive the performance space. As mentioned, Siegfried Linkwitz's ideas say that such an illusion is possible.

This post doesn't even present the argument for this (I'll try not to put words in SL's mouth). It discusses only how I came to question one of my own objections to the argument.
Darren

[darrenyeats]

and listening to a recording of Nathan Milstein play Bach's Chaconne on a Sony clock radio.

Right on!

[youngho]

Darren, I hope you won't mind a few more comments from an amateur:

You probably picked 1m after reading this: http://www.linkwitzlab.com/stereo%20reproduction.htm. 1m is about 3.28 feet. For the rear wall, this represents the the minimum distance from the back wall so that the back wave of the dipole reflects and arrives at the listening position almost 6 milliseconds after the front wave, assuming the speed of sound is approximately 1130 feet/sec. The same applies for an omnipolar speaker. Probably for simplicity's sake, Linkwitz uses 6 msec to represent the minimum delay from the direct signal so that the listener's brain minimizes the reflections' effects relative to the original signal. My understanding of the precedence effect had been that it was much longer than 6 ms, although the degree to which this occurs depends on the levels or amplitude of the reflections. Incidentally, notice how the omnidirectional monopoles are placed closer to the listening position and further from the side walls compared with the dipoles in this particular presentation. A key assumption is that the "reflections are...attenuated copies of the direct sound in spectral content." Again, the quality of reflections matters.

Classical music is not recorded with a single microphone, and Linkwitz discusses the issues of reproducing recorded music at home here (http://www.linkwitzlab.com/reproduction.htm). In short, there are lots of microphones. The input from these are mixed down to 2 channels for stereo. When you play it back on your speakers, it's really not a matter of trying to hear what the microphone hears because there's not one microphone. Even if you could, you might not want to hear what the microphone hears because it may be extremely close to the singer or instrument in question. You'll notice on this page I linked that Linkwitz talks about the illusion of "being there" being strongest when he's close to the speakers as he hears more of the direct signal relative to the reflections, and he talks about the illusion of "they are here" getting stronger when he moves further away from the speakers.

I didn't say that "listening room reflections MUST mean the performance is heard as if in the room" (emphasis mine), but it's clear that there's a gradient where more room reflections means more "they are here" than "being there," even in Linkwitz's stated experience. Even in the first presentation, being able to "hear through" the listening space requires the listener to be relatively close to the speakers (point A), and with dipolar and omnipolar speakers, there are some constraints that I've discussed before that may not be applicable to most listeners.

Ultimately, because for most box speakers the spectral content of the reflections differs from the direct signal, absorption of the first reflection points may reduce the detrimental effects of what Linkwitz calls "confusing cues" on the listening experience. Avoiding too much more absorption will allow late reflections to contribute to the perception of "spaciousness" and enhance the listening experience.

[bpape]

My point with cymbals is that they sound predominantly in a part of the frequency spectrum where listeners are most sensitive to directionality. So, if directionality is a consideration with voices or some other instruments you mentioned, it must be even more so with cymbals.
Darren

True.  Even more reason to kill the off axis reflections because they are not reproduced with the same response curve as the on and close to on-axis response - therefore changing the spectral balance.  Also, in large venue reproduction, there are no close reflections to reinforce those higher frequencies and air is a great absorber (as is the audience).  If we really want to reproduce those as one would hear them in the hall, we reproduce them with the same lack of reflections that the mic heard.

Bryan

[darrenyeats]

Bryan, Youngo,
Thanks again for your replies which I found very useful.

I think our views seem to agree a lot.

Bryan, about the cymbals, yes I see it that way too - with box speakers. In fact, I agree that absorption of reflections is a strategy which is more useful with box speakers because the off-axis response is different. And, I agree with Youngho that this might be limited to first reflections for reasons of air and spaceousness.

Classical music is not recorded with a single microphone

Well, in general. I have some classical recordings done with what I believe is a simple stereo mic in a church...and they sound amazing. I am against multi-mic recordings generally because if you want to re-create an accurate performance space in your living room they kind of spoil the whole endeavor. Of course you can re-create "a" space with multi-mic recordings - no doubt about that. For me, it makes sense to optimise the playback system and room with an assumption of a good (simple stereo mic) recording. Then the effect from multi-mic recordings can fall where it wants...and incidentally IME it comes out very well too!

The SL stuff is not only an academic argument for me. I have listened to the Orions, and on one recording in particular the sound was absolutely real to me. (I have experienced this only twice in my life with sound reproduction.) This recording happened to be from a well-recorded acoustic guitar concert, which only reinforces my belief about simple recordings. The listening room was a totally untreated living room yet the sound was at odds with my vision...it sounded like I was on a stage in a large music hall in Manchester. The illusion of the (untreated) listening space and boundaries dissolving completely was so strong I found it actually disturbing to keep my eyes open.

So I have some subjective reasons for believing SL is on to something. My personal experience is no proof of course. Maybe I heard what I wanted to hear - and also I recognise that there are different approaches which may arrive at similarly excellent results. But this experience is why, personally speaking, I am searching for explanations (I am not there yet, and I thank you both for your continued thoughtful replies).
Darren

[Ethan Winer]

that is not what microphones and stereo speakers do. Instead they record the effect of reflections at the microphone position in the performance space, then re-broadcast the effect from just two points in the listening space. The perception of the listening space is an illusion, not a literal reconstruction.

I agree with this. All the ambience and reverb etc you want was already captured by the microphones. (Or for pop music, added artificially by the engineers.) The job of loudspeakers is to reproduce that cleanly without interference from early reflections in the much smaller room the speakers are in. I'm sure we all agree that a concert recording heard through headphones can sound huge. So that alone proves that reflections in the listening room are not needed for a realistic sounding and satisfying experience.

--Ethan

[TomW16]

I have read that that a completely dead room is not appropriate but at the same time, any reflections in the listening room are to be avoided.  Are these not mutually exclusive goals?  Are we talking about primary reflections only?

Thanks.

Tom

[bpape]

I'd agree that some of the best recordings I've heard are simple 2 mic stereo setups.  They accurately capture what was 'heard' in that point in the hall.

Tom,

A dead room is not pleasant to listen in.  As Scott F. accurately put it - "it sounds like body sized headphones."  The idea is to squelch reflections that are not accurate response wise and are relatively close in time to the direct arriving signal.  We also deal with reflections which can cause bass anomolies.  This is yet another place where large and small spaces differ.  In large spaces, we WANT to build up a reverberant field in the bass - where in small rooms, we need to tame it.

Bryan

[TomW16]

A dead room is not pleasant to listen in.  As Scott F. accurately put it - "it sounds like body sized headphones."  The idea is to squelch reflections that are not accurate response wise and are relatively close in time to the direct arriving signal.  We also deal with reflections which can cause bass anomolies.  This is yet another place where large and small spaces differ.  In large spaces, we WANT to build up a reverberant field in the bass - where in small rooms, we need to tame it.

Bryan

I think I understand what you are saying; for small rooms (what I am interested in), first reflections at the listener position should be mitigated as much as possible.  In addition, bass frequency anomolies need to be corrected, typically with bass traping. 

Secondary reflections that reach the listening position are not as critical presumably because they are not as close in time to the original transmission and because they are not as accurate to the original transmission and, therefore, easier for the brain to filter out.

Let me know if I am off here.

Thanks.
Tom

[darrenyeats]

that is not what microphones and stereo speakers do. Instead they record the effect of reflections at the microphone position in the performance space, then re-broadcast the effect from just two points in the listening space. The perception of the listening space is an illusion, not a literal reconstruction.

I agree with this.

Ethan, I did mean to write "performance space"...I'm sure this is what you thought I meant too.
Darren

[giantsteps]

 Here's my carefully planned SOTA rear wall. 

 
 


 Frank

[bpape]

Tom. 

That's basically correct.  As an example, the direct reflection off the rear wall (mid and high frequencies) are mostly from relatively on-axis of the speaker so they are fairly accurate in their spectral balance, unlike direct side wall reflections which are very skewed in the high end in most cases. 

Also, the ones off the rear (assuming you're sitting appropriately far from the wall behind you) are much later arriving and can give a sense of spaciousness to the sound.  Let's just look at a hypothetical situation.

I'm in a 12x15 room sitting approx 5' from the wall behind me.  The speaker faces are 3' from the front wall.  The direct path to me is approx 7.5 feet.  The side wall reflected path is going to be (off the top of my head) around 11' assuming the speakers are a couple feet off the wall.  The path for the rear reflections is going to be approx 17'.

As a general rule of thumb to read other postings and theory, you can use 1' = 1ms in time.  So, the side wall reflections which are already spectrally unbalanced are going to reach me within approx 4ms of the direct sound.  The rear reflections will reach me approx 10ms after the direct sound.

Bryan

[youngho]

I have to admit that I have very few simply miked recordings. The ones I have, which are mostly jazz and acoustic guitar (for example, Minton's Playhouse, Luis Bonfa), don't capture a lot of the hall ambiance but rather have an intimate presentation. My EXTREMELY limited understanding was that single stereo microphone recordings are relatively close to mono because of the minimal separation and that 2 microphone stereo recordings can sound fantastic but may capture much more of the reflections and reverberant field (and audience noise if one is present, or else the absence of an audience results in a somewhat different hall acoustic than if one were present) relative to the direct signal, which may be pleasing to the ear (not just "you are there" but more so).

Frank, Floyd Toole mentions the use of a pillow behind the head as a quick test for whether rear wall absorption may be helpful!

[Ethan Winer]

Frank, Floyd Toole mentions the use of a pillow behind the head as a quick test for whether rear wall absorption may be helpful!

I have an even quicker "test" - if the wall behind you is ten feet away or closer, you will benefit from absorption there.

--Ethan