Very clever point!
At some point way down the hall, maybe it starts to sound indistinguishable.... or after you've had a lot of alcohol to drink.
I think the closest I've come is playing cello music on a Karlson-15 (say, 1957) which kind of looks like a cello and is all woody and resonant.
Perhaps a single speaker (maybe even a single OB driver???*) can do the trick better than a stereo pair. Ummm?
Oddly, I think it is easier to close your eyes and think "the real instrument is in my room" with organ music because the source in the church is very diffuse** unlike trying to make two flat speakers sound like one tubular flute.
I gotta laugh (but respectfully) at the high-math that goes into stereo analysis. What that math is aimed at is not making a flute play in your room but making your speakers sound like "real" speakers. See my next post.
*Oh gosh, what might I be starting with that thought? But the logical conclusion might be inescapable that ESL-dipoles and OBs rule! Or headphones, eh.
** Organ music in concert halls always sounds a bit like a bad reproduction of church organ music. Which supports my point.
Last edited:
Are you kidding or are you referring the famous tests conducted by that guy from New Jersey, Tom Alva something, with the wax cylinders?
But the smart guy is Vilchur. His live-versus-recroded tests were making an AR3 indistinguishable from itself! That is a smart test, it is feasible, and although a rather modest goal, it all you can do since you can't make a flute wave-front using two three-way speakers pushing sound in your face whatever their math-sanctified directivity index.
Last edited:
Your valuable long post discusses the two sides of psycho-acoustics. On the acoustics side, your chart shows some factors which I think it is fair to call "disruptive" to those souls who hold to a simple model of stereo loudness. It deals with the physics of sound albeit as received by meat heads.
The "psycho-" side comes up when you address how irregular/non-linear loudness perception is. Just looking at loudness of sound per se, it is influenced by tonal context, time course, frequency, etc. and we are far from seeing how that works in-vitro in the lab let alone when music is playing and we are cranking higher our bass.
Well my tongue is in my cheek; the main point being that the test is invalid because expectancy overwhelmes information in live vs loudspeaker. The only way to assess this would be blind. Some blind tests suggest it is plausible once you get over 20 feet from the performers and loudspeakers (Talk:High fidelity - Wikipedia, the free encyclopedia), and the whole approach is not useful as a means of evaluating loudspeakers (Audio Musings by Sean Olive: Why Live-versus-Recorded Listening Tests Don't Work).
I'm not sure that Edison ever did those tests with his cylinders, I'd have to see documentation of that. But he did extensive live vs recorded demos with his Diamond Disc players. If you've ever heard one in good playing trim* - you'll not sneer at the tests any more. They are amazing. Add to that the fact that Edison taught the singers to sing like the recording (for the demos) and it's all quite plausible.
*These are old devices now, and need repair and new parts.
Let's go back to Sanedesign's post #558, that started all this "real- not real" silliness.
I don't see anything in there about Carnegie Hall or any other concert hall. Unless you read "live performance" as concert hall. When I read that it struck a chord, but I wasn't thinking genuine hall acoustics. I've only head one (1) system that could really do that. Does it sound like a real human voice, a violin, a contrabass? Or does it sound like a recording of such? To me, that's the point.
And that is what first struck me with very high end systems so many decades ago. "That sounds real!" Not "that sounds good." I go to a lot of Hi-Fi shows and hear many systems - big and small - that sound "good". But very few sound "real". Very few sound like real music. To me, that's the point.
I don't see anything in there about Carnegie Hall or any other concert hall. Unless you read "live performance" as concert hall. When I read that it struck a chord, but I wasn't thinking genuine hall acoustics. I've only head one (1) system that could really do that. Does it sound like a real human voice, a violin, a contrabass? Or does it sound like a recording of such? To me, that's the point.
And that is what first struck me with very high end systems so many decades ago. "That sounds real!" Not "that sounds good." I go to a lot of Hi-Fi shows and hear many systems - big and small - that sound "good". But very few sound "real". Very few sound like real music. To me, that's the point.
Lots of speakers, channels, and EQ should work, if the hallway has enough openings and length. Might even work in room.
In your general life, walking through the store, the woods, speaking with your loved ones, do you perceive large changes in tonal balance as you rotate your head or move position. Of course not. Your perception doesn't end at the ear drum. You know you're turning your head.
To illustrate this, take a live dummy head binaural mic feed and play it into a pair of headphones on your head. Have someone walk around the dummy head and speak. You do hear the tonality change in a way that is quite unatural. I've enjoyed this demo for myself.
Your premise would imply that human tonal perception is immune to HRTF changes over angle. That would be an interesting premise to prove, I'm "all ears".
Dave
I haven't compared stereo to mono but I would definitely say that the OB speaker (or ESL likely although I don't own one.... yet) sounds much more real _especially_ when listening from another room when compared to any other "boxed" speaker I've heard and this would be at any listening position and especially from another room or down the hall.
Although I do have a digital piano in my living room, perhaps it would be a bit of a stretch to say it "tricked me" into thinking there was a real acoustic piano in my living room.
However my OB speakers _have_ tricked my wife into thinking I was playing something on my real acoustic guitar that I also have in the living room.
Last edited:
So in one short sweet anecdotal test you've invalidated the whole premise ? If only it were so easy 😉 With speakers of what directivity profile, in what type of room ? Listening in the near field or in the reverberant field ? Have you tried the same test outdoors ?
I'll give you a counter example. If I listen to a system with a full range driver crossing over to a ribbon tweeter at ~4Khz in a "normal" listening room with a significant amount of reflection/reverberation, and turn my head left and right approx 30 degrees as you describe, I notice almost no change in tonal balance and the phantom image continues to stay centred between the speakers, just as you suggest.
However if I listen to the same full range drivers by themselves, (driven full range) which go up to 15Khz but are very narrow in the treble and then perform the same test, there is a dramatic difference in perceived treble with head rotation.
The ear that turns towards the speakers gets a big boost in treble while the ear that turns away loses a lot of treble. It's not a subtle effect, it's very obvious, and it affects both the tonal balance and phantom centre channel image.
Why the difference ? I think it's reasonable to suggest that when you listen to a "normal" speaker with wide treble dispersion in a normal room your ear is picking up a lot of non-direct treble from the room, and the sound that comes in from a wide listening angle (relative to the centre line) has it's apparent treble greatly exaggerated by the HRTF.
This means the reverberant treble field you're hearing from the "sides" of the room can rival or dominate the direct treble from the speakers, even though an omni-mic measurement made at that location would show the direct treble to be stronger.
In other words the HRTF has greatly reduced the apparent direct to reflected ratio in the treble region.
In the case of the driver with very directional beam like treble, there is almost no treble in the reverberant field so as soon as you start to turn away from the speakers the HRTF of the ear turning away reduces the treble and the HRTF of the ear turning towards the speakers boosts the treble - and there is no swamping effect from reverberant treble to "hide" this change in the direct field.
Well, I feel that DDF is right about the fact that there is a difference in treble balance between a single mono speaker directly ahead and two stereo speakers at 30 degrees producing a phantom mono image, due to the HRTF of our ears. It's pretty hard to deny when it can be measured with dummy heads.
However, I don't agree that its the job of the speaker designer to attempt to compensate for the effect, and that in fact it's already taken care of in most cases.
Whenever a stereo recording is mixed it will be tested on stereo speaker pairs at ~30 degrees, and for this reason I believe that nearly all mixing engineers will be knowingly or (probably) unknowingly compensating for it in their mix by basically shelving the treble down a bit if it sounds too bright.
They might not know exactly why it sounds too bright, and there will be other factors that will be affecting how bright it sounds, but at the end of the day the final compensation curve will include this effect if it is to sound balanced.
It does mean that simple pan-potting of sources will affect their perceived tonal balance slightly (brighter than it should be when panned to the middle) but again this is almost certainly taken care of in the mix if each channel has it's own EQ.
As long as the mixing/mastering reference is stereo speaker pairs, and the playback reference is stereo pairs, and EQ'ing is done during mixing, there is no real issue.
There are only three cases I can think of where this issue would need to be consciously addressed:
1) Doing a live unmixed un-EQ'ed recording of for example a concert performance. This is likely to sound too bright both because of the 30 degree HRTF treble gain on the playback speakers discussed here, as well as cardioid microphones capturing "too much" ambient treble compared to human ears. Some amount of shelving down of treble is probably necessary here, or perhaps doing it as a binaural recording with a dummy head.
2) Listening to an old, genuine mono recording (early Beatles?) which was mastered and mixed with a single mono speaker in the mixing booth - in this case the treble would be too bright listened to as a phantom mono signal on a stereo pair. However if the recording was mixed or remixed (and re-EQ'ed) in mono on a stereo monitor pair, the effect would be taken care of.
3) Listening to a 2 channel stereo recording with a 3 channel (Left-Centre-Right) speaker setup - as part of "decoding" the stereo signal into a Left, Centre and Right channel, the decoder needs to add a small boost in treble to the signal sent to the centre channel, otherwise the centre will sound dull.
Conversely, listening to a 3 (or more) channel mix on a stereo speaker pair, for example playing a 5.1 DVD with a centre dialogue channel on a 2 speaker system, the centre channel (only) would need to have a small cut to the treble before being sent to Left and Right - maybe some processors do this, maybe some don't.
These are unusual cases though, for straight 2 channel mixing/mastering and playback, I think the effect has been inherently compensated for all along...
The one time I was surprised was at the stereo shop I frequented years ago. I walked up to the main counter to speak to the owner, a friend of mine, and did a head-turn double take. In a smallish demo room with the door open someone had just started playing a cut from the MTV Uplugged recording of Eric Clapton. I wasn't paying attention to the intro, but when he began to sing it made me turn my head to the room because it had an uncanny realism. This was on a set of B&W 802s on the left wall facing into the room. I was essentially entirely in the off-axis reverberant area. I couldn't even see the speakers from my position.
The only other anecdotal remarks I care to submit is that two of my three dogs were lying on the floor when I put on a recording from the original Circle album that includes pre-song recording room banter in my current dipole effort. They both picked up their heads and looked at a point roughly between the speakers with one of them rotating his head back and forth as dogs (and people) do to attempt to localize an object, such as a bird in a tree. It didn't last long, of course. Obviously they were briefly fooled into thinking that someone had begun to speak in the room.
That amount of realism is generally enough for me.
Dave
thanks for helping DBMandrake....
I would agree there is going to be _some_ inherent compensation in the mastering as you have eloquently described.... but my suspicion is that most of living rooms have have a sonic characteristic such that it would most often benefit from some additional compensation.
Even JohnK provides a sort of shelving HF rolloff circuit on his speakers and he doesn't buy into this "flat is not correct for stereo" theory.
I've addressed your critique of Dave's post in my previous reply, but I think you've still missed the entire thrust of my initial post.
I wasn't trying to provide more evidence for Dave's "30 degree HRTF error" issue, but rather, reading it got me thinking along different lines about another possible reason why "flat and wide dispersion" sounds too bright.
In other words, the increasing high frequency sensitivity of the ear towards 90 degrees (relative to either directly forward or even 30 degrees for that matter) due to HRTF exaggerates the treble of the reflected sound that enters the ear from wide angles such as the first early side wall reflections at ~60 degrees, or lateral reverberation in general.
This is the first explanation I've seen that could reasonably account for why a flat power response in the treble in a room sounds too bright when listened to with human ears...Lots of studies have concluded that a power response that tapers down somewhat in the treble sounds best in a reverberant room but nobody seems to know or want to say why, other than that it is.
I agree that a flat on-axis response is the primary determining factor for perceived balance, (>200Hz) however this only holds true if the on-axis sound is significantly louder (at the eardrum) than the reverberant field. If the reverberant field is relatively strong, then the total perceived balance becomes affected by it.
If the HRTF of the ears can boost treble coming in from wide angles dramatically, it may reduce the direct to reflected ratio (as far as our ear drums are concerned) by so much that the reverberant field becomes a significant determining factor for tonal balance in the treble.
Not sure what you're trying to get at here - I'm not familiar with the specifics of either studio monitor, I'm just relating what I've noticed with different directivity profiles and how they interact with different rooms.
It seems pretty clear to me that very wide dispersion at high frequencies is the wrong approach, and I've thought that for many many years. Very directional is no good either, there is some middle ground that gives the best results.
Aiming for a flat on-axis response in the treble and modifying the directivity vs frequency profile of the tweeter seems to be the right approach to getting a speaker that will both measure good and sound good, and have minimal variations in perceived treble balance with different room acoustics.
The big advantage of a wave guide tweeter over a typical dome is that it controls the low/mid treble "bloom" in the power response that you would otherwise get with a typical dome tweeter crossed over to a cone midrange driver. Funnily enough this bloom occurs in exactly the region where the ear becomes more sensitive with increasing angle - from the tweeter crossover up to about 8Khz.
As well as controlling the dispersion in the low/mid treble a wave guide design can more closely match the dispersion of the midrange driver at crossover minimizing abrupt changes in power response, reduce cabinet diffraction effects in the treble, (since little high frequency energy is travelling across the baffle surface) provide a deeper acoustic centre to minimize or eliminate differences in acoustic centre between the tweeter and midrange, and improve sensitivity and distortion figures due to a small amount of horn loading.
I really do think the current typical approach of trying to make a tweeters dispersion as wide as possible at higher frequencies is wrong. Instead we should be trying to control the directivity at the lower end of the tweeters range for all of the above reasons.
Many ribbon tweeters use wave-guide slot loading out of necessity to provide extra sensitivity, lower distortion, and to recess the ribbon to protect it from damage, but maybe one of the reasons why ribbons can sound so good is precisely because of their directivity control.
You may be right, I honestly don't know. I guess I have less faith in teh recording process than you do. This is the classic conundrum of "references" again.
However, this still wouldn't address the difference in HRTF between you and those that made the recording. This in itself could explain some of why different people hear different tonal balance from the same system (mostly though, the differences in qualitative tonal perception probably comes down mostly to what they live with day-day). In the grand scheme of recording/system errors, this difference is probably small fish though.
Getting back to tnargs argument, if the frequency response to the DRP doesn't matter, than neither does the frequency response of the stereo itself. We of course do know that the frequency response matters, and knowing that you're rotating your head matters as well.
Dave
Sane, I think you are right on with this post though you stated it too softly. There obviously is compensation in these recordings provided the engineers are not deaf(I'm sure some of them are). Even then, check out the El Greco Fallacy. Just look at how my room and Pano's room started out. The question to me is "do we fix time errors in the time domain, or do we fix them in the frequency domain?" I'm sure we all know my position--you cannot fix acoustic problems with an EQ. Arguments could be made either way, but I don't want to 'fix' a problem by creating another even though it is really easy to just turn down the treble. Knobs are great.
Fact is, the better my room and speakers have measured(FLAT of course), the better the tonal balance and enjoyment of my recordings have been.
Dan
If something is going to sound "real" (like real music), then the system has to produce an acoustic environment like (but not necessarily exactly like) that of the original performance.
I think the only way to do that is replace the home music room's acoustic with one like that of the original performance.
I'm sure you can't do that with stereo just because you'll need more than two speakers.
I figure extreme modification of listening room in direction of anechoic is not necessary. What's needed is enough point sources with enough of the right kinds of EQ and delays to modify the existing acoustic environment
From my very modest experiments a while ago I found I could get close to "real" just using home theatre receiver, dcx 2496 and some old integrated amps.
The downside was adjustments had to be made from recording to recording depending on what sort of music it was. I listen to classical music mostly, and I suspect it takes kindly to this sort of treatment.
When I get some more small speakers I'll play with it in a more systematic fashion.
I think the only way to do that is replace the home music room's acoustic with one like that of the original performance.
I'm sure you can't do that with stereo just because you'll need more than two speakers.
I figure extreme modification of listening room in direction of anechoic is not necessary. What's needed is enough point sources with enough of the right kinds of EQ and delays to modify the existing acoustic environment
From my very modest experiments a while ago I found I could get close to "real" just using home theatre receiver, dcx 2496 and some old integrated amps.
The downside was adjustments had to be made from recording to recording depending on what sort of music it was. I listen to classical music mostly, and I suspect it takes kindly to this sort of treatment.
When I get some more small speakers I'll play with it in a more systematic fashion.
It's not that I have a lot of faith in the recording engineers (for many recordings I don't) it's just that it's something that more often than not would be implicitly corrected for in the mix, even if incidentally (provided the engineer makes the effort to EQ the mix to sound "right", and is skilled enough to make the right adjustments...) and that by trying to put a "correction" in the speakers response you run the risk of applying the same corrective action twice - once implicitly in the mixing of the recording, and a second time in the speaker response resulting in a recording that is now lacking in treble, which although easier and less grating to listen to, is just as "wrong" as a recording that has too much treble.
For this scenario to work the mixing reference speakers would also have to be "corrected" the same amount as the end users speakers, and I really can't see this happening, if only because it would cause an apparent error in the vast majority of already mastered material.
No, I think we're stuck with the current reference situation where "flat" recorded signal in the stereo format implies an increased acoustic treble response due to HRTF, at both the mastering and playback stages.
Where it does need explicit correction is converting 2 channel material for 3 channel (L-C-R) playback and vica versa, but then it's the job of the signal processor to do so, not the speaker.
Also anyone trying to make a live un-mixed recording needs to be aware of the fact that flat will probably sound too bright when played back, so at the very least some adjustment of the treble may be necessary. They probably already do this, but may not know why it's needed.
Good point, not only is it probably one source of different perception of tonal balance between different people in stereo listening, (for example between the mixing engineer and the end listener) the induced error couldn't be corrected for exactly anyway because not only are everyone's ears different with their own unique HRTF, not all listeners position their speakers 30 degrees to either side, either due to lack of positioning options in the room, different tastes, or in the case of the average non-audiophile listener, a lack of appreciation of the importance of speaker positioning.
For example my speakers are only about 24 degrees to either side, that's the widest I can get them due to other room objects, but when I have moved things to set them a bit wider I find I prefer them where they are anyway - at least in this room.
It probably is small in that sense, although still important, particularly when converting from one number of channels to another. One thing is clear though, it's just yet another reason why "flat" (right from microphone to speakers) may not necessarily be correct in stereo sound reproduction, as if more reasons were needed 😛
The thing is...
Anyone who's tried to do high quality recordings of acoustic instruments will know that any recording process has limitations and complications, and it's not easy to make a recording that will create a very perfect reproduction through two speakers in a typical listening room. So as any experienced recording engineer is likely to know, you "sweeten" the recording based on a prediction of what will work well in the typical playback situation (maybe add reverb, EQ, compression, de-essing, etc.), or go for flat/neutral, maybe for the sake of consistency, and hope the end user of the recording will have reasonable acoustics, speakers and tone controls. It's never going to be "perfect" or consistent from one recording to another. That's not possible with two speakers. What is perfect? What's the goal here? I say it's to have a great experience with an audio program. Some recordings played back through my system sound way better than I've ever heard live, and some don't. It's all over the place. Because of that, I'm a very firm believer in having a certain style of tone controls (four section variable slope). When rich people buy those >$1K preamps (some are >$25K) that have only an input selector switch and a volume control, all I can do is laugh. It's about enjoyment, not someone's definition of technical accuracy, and accuracy that isn't really fully possible anyway.🙁
Anyone who's tried to do high quality recordings of acoustic instruments will know that any recording process has limitations and complications, and it's not easy to make a recording that will create a very perfect reproduction through two speakers in a typical listening room. So as any experienced recording engineer is likely to know, you "sweeten" the recording based on a prediction of what will work well in the typical playback situation (maybe add reverb, EQ, compression, de-essing, etc.), or go for flat/neutral, maybe for the sake of consistency, and hope the end user of the recording will have reasonable acoustics, speakers and tone controls. It's never going to be "perfect" or consistent from one recording to another. That's not possible with two speakers. What is perfect? What's the goal here? I say it's to have a great experience with an audio program. Some recordings played back through my system sound way better than I've ever heard live, and some don't. It's all over the place. Because of that, I'm a very firm believer in having a certain style of tone controls (four section variable slope). When rich people buy those >$1K preamps (some are >$25K) that have only an input selector switch and a volume control, all I can do is laugh. It's about enjoyment, not someone's definition of technical accuracy, and accuracy that isn't really fully possible anyway.🙁
Firstly, thanks for the strawman argument, picking on my simple *illustrative* example and pretending it was the only evidence in the world. Then producing a long post of your *personal, sighted experiments* and your conclusions from those tests about how psychoacoustics works! If you want to know about this stuff then I suggest Clark and maybe Perrot in the 80's , Shirley in the last few years, and Watkins and I think Makin (? somebody help me here) in the 90's. Forget the "lounge room science" you have been conducting.
No need to argue the errors in the phantom stereo image; they are proven and I referred to their physical reality in this thread a few days ago. But to claim it is compensated for by nearly all mixing engineers "knowingly or (probably) unknowingly" is (a) wrong, and (b) impossible to do well if you tried. The resultant mix would only be right at (a) one angle of listening speakers, (b) one placement of listening speakers in the room, (c) one size and shape of listening room with one type of wall surfaces, one floor treatment and one layout and type of furnishings. Hmm! This merely serves to illustrate the lack of standards and broken circle of reproduction in music recordings.
Your conclusions about reflected sound dominating the direct sound are in utter contradiction to all known good science on this matter.
Therefore your conclusions about how best to select and deploy hifi equipment at home are suspect.
- Status
- Not open for further replies.