Everyone seems to design as if stereo bass is important, but I am skeptical.
Can anyone point me to a scientific study of human sound localization vs. frequency? I am less interested in personal anecdotes of "I can be spun around blindfolded 100 times and shoot a 50Hz tone right between the eyes every time."
I did some concerted Google-fu but came up wanting. The following excerpt from the Wikipedia article on sound localization was somewhat helpful:
That little "citation needed" part is the irksome bit. I imagine 80Hz isn't a brick wall. So how much above 80Hz does the perceptual system begin to lose its grip? If we can localize laterally to within 1 degree at 1kHz, How many degrees of uncertainty have we accumulated down at 150Hz?
And none of this addresses the question of vertical localization of low frequencies where interaural time delay cannot be the mechanism. I seem to remember reading something about the brain estimating low-frequency source elevation by analyzing floor-bounce cancellations...
Can anyone point me to a scientific study of human sound localization vs. frequency? I am less interested in personal anecdotes of "I can be spun around blindfolded 100 times and shoot a 50Hz tone right between the eyes every time."
I did some concerted Google-fu but came up wanting. The following excerpt from the Wikipedia article on sound localization was somewhat helpful:
That little "citation needed" part is the irksome bit. I imagine 80Hz isn't a brick wall. So how much above 80Hz does the perceptual system begin to lose its grip? If we can localize laterally to within 1 degree at 1kHz, How many degrees of uncertainty have we accumulated down at 150Hz?
And none of this addresses the question of vertical localization of low frequencies where interaural time delay cannot be the mechanism. I seem to remember reading something about the brain estimating low-frequency source elevation by analyzing floor-bounce cancellations...
Bill, on the flip side prove to us that it is not. I can do it blindfolded. But if using pure tones it can be hard to locate most sounds. Use band filtered noise and it's much easier. Try it yourself.
I know you said no personal stories, but I want to see solid evidence that it is not before I believe that it is not.
I know you said no personal stories, but I want to see solid evidence that it is not before I believe that it is not.
It is important if you only want to design and build two boxes.
Silliness aside, it may come down to different personal levels of perception. Some, like Pano, can differentiate while others may have a different threshold level. It may well come back to the lower level of difficulty and lower use of floor space to incorporate all of the bass drivers into the stereo boxes.
My question is; why are you asking the question? (As you well know, such open questions often invite comments from people with "odd" ideas, many of whom are on my ignore list.) Do you wish to open a large can-of-worms that will, in all probability, offer nothing more than an exercise in the abstruse.
Someone needs to make a test.
Put sub boxes all over a room. As many as can be built and afforded. Even empty boxes for placebo. Just make them down firing, or place woofer away from listener. Blindfolds even.
You can use a minidsp or something with multiple line outs with volume controls for each sub. Adjust volume of each sub one at a time randomly and have listener point to where they think the sound is coming from.
With minidsp the crossover can be fiddled with. So the crossover point can be raised and lowered to see what effect it has as crossover is changed.
Put sub boxes all over a room. As many as can be built and afforded. Even empty boxes for placebo. Just make them down firing, or place woofer away from listener. Blindfolds even.
You can use a minidsp or something with multiple line outs with volume controls for each sub. Adjust volume of each sub one at a time randomly and have listener point to where they think the sound is coming from.
With minidsp the crossover can be fiddled with. So the crossover point can be raised and lowered to see what effect it has as crossover is changed.
Yes it can, in free-field or in a large room.
Not so clear in a small room, where modal effects dominate. We can hear 'something' resembling a localization cue, but you could argue it is a false que, due to sitting near a pressure gradient.
Griesinger used to argue for it while he was at Harman, but it was found that a) >99% recordings don't support it and b) if everything is optimized for this effect, it is still marginal at best. Harman went for SFM instead. Contrary to the 'spatial bass' effect, taking care of modal problems is clearly audible on every recording (containing frequencies <200Hz) and the effect is not marginal.
If you have a Lexicon processor, there is a 'bass enhance' function, or you could DIY the same thing: don't sit in the rooms centreline, bass sources 90deg left and right, phase of one source lags the other ~90deg.
Not so clear in a small room, where modal effects dominate. We can hear 'something' resembling a localization cue, but you could argue it is a false que, due to sitting near a pressure gradient.
Griesinger used to argue for it while he was at Harman, but it was found that a) >99% recordings don't support it and b) if everything is optimized for this effect, it is still marginal at best. Harman went for SFM instead. Contrary to the 'spatial bass' effect, taking care of modal problems is clearly audible on every recording (containing frequencies <200Hz) and the effect is not marginal.
If you have a Lexicon processor, there is a 'bass enhance' function, or you could DIY the same thing: don't sit in the rooms centreline, bass sources 90deg left and right, phase of one source lags the other ~90deg.
I think it´s important to keep in mind that these boxes will have resonances and harmonics (distortion) occurring at higher frequencies which will make it easy to localize them. For a real test one would need some source of a pure low frequency.
If it's a musical instrument playing a tone, yes, it can be localised.
Because most instruments generate lots of harmonics. However common music recording techniques concentrate on putting as much of the actual low frequency at the center because of phase issues, though sometimes engineers may choose to leave LF where it actually occurs. For example, the floor toms in Hotel California, or in James Taylor 'Gaia' can only be properly reproduced with full-range speakers on either side (I don't mean one speaker full range, but 20-20k ability with decent SPL)
If it's a subwoofer+main speakers or woofer playing back that tone with it's own distortions, again, yes. Even with the LF portion of the tone emanating only from the center we can pick out the pluck of strings or bow - and the brain tends to localise the sound source - assuming a good room and speakers. It would be impossible to record double bass drums without, and that's just one example.
A pure single 50Hz in an anechoic environment - maybe, but then is that a scientific question rather than relevant to real speakers in real rooms playing back music?
Because most instruments generate lots of harmonics. However common music recording techniques concentrate on putting as much of the actual low frequency at the center because of phase issues, though sometimes engineers may choose to leave LF where it actually occurs. For example, the floor toms in Hotel California, or in James Taylor 'Gaia' can only be properly reproduced with full-range speakers on either side (I don't mean one speaker full range, but 20-20k ability with decent SPL)
If it's a subwoofer+main speakers or woofer playing back that tone with it's own distortions, again, yes. Even with the LF portion of the tone emanating only from the center we can pick out the pluck of strings or bow - and the brain tends to localise the sound source - assuming a good room and speakers. It would be impossible to record double bass drums without, and that's just one example.
A pure single 50Hz in an anechoic environment - maybe, but then is that a scientific question rather than relevant to real speakers in real rooms playing back music?
Last edited:
Okay fine, Pano, YOU are allowed an anecdote.
Merry Christmas.Myself, I'm sorry to disappoint, but I will not be providing any useful evidence one way or another. A useful result must exclude any mechanical or port noise, any box talk, any higher harmonic content, etc., that would provide undue cues for localization, and I know my limits.
My listening "laboratory" is a 110-year-old house with squeaky wood floors and rattling windows. Even if I had 10 bass boxes I could arrange into a 180-degree arc and a helper running a switch box for a blind test, there would be too many induced squeaks and rattles coming from every which way. That's why I'm interested in the results of those who have the wherewithal to perform legitimate scientific inquiry.
There must be a paper or two out there with good methodology and definitive answers...
Very true.
Regarding floor bounce cancellations, sorry I wasn't clear: I wasn't talking about bass, but higher frequencies where floor bounce would manifest, but frequencies too low for pinna transform to provide an elevation cue. What I recall reading is that human aural processing can use floor cancellation cues to assign an elevation to a sound source. But I can't find the study again...
Ah! So you are skeptical of my skepticism. Well played sir!
Here's the background to why I'm asking the question:
My living room is rather typically sized and has a sparse modal zone (problem zone!) from about 40Hz to 120Hz. Thinking of how this could be overcome in an ideal scenario led me to the technically sweet "double bass array" (DBA) concept that attempts to launch plane waves the size of your entire front wall. Of course, this setup requires summing to mono--up to 120Hz in my case. Would I be giving up anything important in losing stereo separation up to 120Hz? I expect my perception would still be dominated by the properly located mid and treble cues, but this is just my conjecture.
Also, I thought this might be a useful discussion since co-locating bass sources with mid and treble may not provide the smoothest bass response. People seem to assume this is where bass speakers MUST be located, and this is common sense, but I'm just wondering if it has a solid perceptual basis, and if that justifies lumpier bass, which I think we can all agree is very audible.
If it turns out we have more perceptual wiggle room than we thought we did, then we can feel encouraged to seek higher levels of system refinement by moving our bass cabs around.
Last edited:
The lower registers where the male voice predominates is normally associated with the woofer in a 2-3 way so placing them together was (is) the only way for proper integration. If we break the traditional mold, and place midbass/woofer/sub into their appropriate zones we can and often do move the subs or like per Geddes multi point/flank them. To get all that right limits the system to one sweet spot, but at least we've got one
The first system I built to address this was a sub sat system back in '84. Limits on technology in that day held back what could be achieved eg digital delay in 84? not happening
Bill,
1) Your conjecture is correct.
2) Most (good) recordings have bass instruments centered, so there is little stereo separation below 120 Hz to start with.
3) Although I have no problem locating the origin of audio below 120 Hz outdoors, it is more difficult in small rooms.
If the DBA does indeed eliminate room modes below around 120 Hz, the advantages in sound quality would outweigh the lack of the few directional cues that would no longer exist that low.
Art
Obviously much depends on where you split your bands. In a typical 3-way, the woofer might go up to 400Hz or more, and it could certainly be audibly localized (in azimuth) in its upper range by the perceptual mechanism of interaural time delay (ITD).
But I'd say it's far less of a slam dunk at lower freqs. ITD becomes less useful as ear-to-ear phase difference shrinks. At what point does the perceptual system begin to lose its grip, and what is the range of uncertainty? This is what I would like to find out.
Also, I believe Dr. Geddes would argue the opposite about multi-subs. They tend to average out room response lumps for smoother response over a wide area. If all you care about is the bass response at a single seat, then a single sub and an hour or two of move-and-measure can deliver a decent result.
Last edited:
Yes, there has been research done in this area. However, relying on Wiki or google will not get you very far. So far the question needs a bit more defintion before it can be answered.
Certainly down at 200 Hz, there is still decent localization. The mechanism would not be binaural differences in intesnsity, but rather binaural differences in time (it is more productive to think in units of time rather than phase).
Certainly down at 200 Hz, there is still decent localization. The mechanism would not be binaural differences in intesnsity, but rather binaural differences in time (it is more productive to think in units of time rather than phase).
Couple issues here-
1) Band-limited noise- perhaps higher frequency intermodulation artifacts are what you're localizing with band-limited.
2) I want to see solid evidence that your localization is not related to higher frequency artifacts before I believe it is not (note the fallacy of disproof).
I like lots of Sd in subs, finding that big subs disappear better- less localization. This would tend to indicate that the localization is tied to a higher frequency distortion mechanism. I get the same reduction in location cues from proper isobaric mounting, or bandpass mechanisms. I do run my mains full range with 40-50hz extension, so it's less an issue for me than someone considering a stereo setup with 80-120hz mains HP and trying to decide between mono or stereo subs.
Last edited:
My dear W/Tarragon, if we cannot expect Wikipedia and Google to deliver the known universe into our laps, then are we not a lost generation????
Let me try to reiterate my question:
Does anyone know of a scientific study quantifying the average human ability to locate sound sources--in azimuth or in elevation--over the range of frequencies between, say, 50Hz and 200Hz?
I have already mentioned the mechanisms of interaural time delay (azimuthal localization) and floor-cancellation analysis (elevational localization). There may be more, so feel free to shed any new light. What remains to be established is exactly how well these localization mechanisms work at the frequencies in question.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.