Have you read Griesinger?
Inner ear behaves live spectrum analyzer with coupled processing for harmonics, and primary detector has rectifier behavior.
Wiring of locational processing is primal in evolution for survival. Short wavelength take least time to process, but don't travel well.
In time it takes higher thought process to put an experience to a received sound, like a twig snapping under foot, the primal processing has already determined direction, and even size of source. This is all clearly observed with startle reflex.
Regards,
Andrew
All impulsive sounds start at t=0, amplitude=0; thus all frequency components are in phase at least once during sound formation. Harmonic content is shorter wavelength/processing time, and relative levels and experience are enough to know original source has lower fundamental.
Thus Griesinger highlights importance of about 700hz to 4kHz in source identification and localization. Adequate information for impulsive sources of these frequencies is decoded in a few hundred microseconds to about 2ms.
Reflections occurring during or shortly after this period add to original impulse, and modify the perception of size, loudness, and location.
With my speaker placed four feet from wall, early reflection is floor, and is greater than 3.5ms for nearfield listening. Direct sound is clean, and brain has no trouble forming directional picture from these, and rejecting later reflections for ascertaining direction.
The rest is brain's s susceptibility to illusion created by highly correlated signals from multiple sources.
That's really interesting, I need to read more. A sort of coincidence for me, as that very close to the range of my midrange horns.
Griesinger is a very knowledgeable guy but he also tends to make his hypotheses sound like proven facts. He's also most of the time talking about perception of real sources. Stereo is different.
Our hearing doesn't simply "reject" reflections. It can be easily shown how later arriving reflections override the direct signal. The performance of an omni will depend heavily on the room hence making it rather unpredictable.
Fact is that an omni creates lots of room reflections and the effects you've described are not "completely separable from room reflections". The listening experience of your omni in the near field vs. far field should be very different because of room reflections.
By the way, a Pluto type of construction creates diffraction too. It's probably just not so easy to measure.
Our hearing doesn't simply "reject" reflections. It can be easily shown how later arriving reflections override the direct signal. The performance of an omni will depend heavily on the room hence making it rather unpredictable.
Fact is that an omni creates lots of room reflections and the effects you've described are not "completely separable from room reflections". The listening experience of your omni in the near field vs. far field should be very different because of room reflections.
By the way, a Pluto type of construction creates diffraction too. It's probably just not so easy to measure.
Last edited:
Precedence effect with respect to early timing cues is rock solid easy to demonstrate stuff. Shockingly Pluto type speaker images just as well when placed three feet in front of brickwall, or outdoors, or at 4 meters, etc.
When one speaker is placed near side wall and other isn't, imaging get lopsided.
There is no diffraction pattern from two drivers due to close spacing v crossover frequency. 1/4 to 1/3 wave ripple over 90 degrees. No lobes.
Regards,
Andrew
Hey Andrew, what is the difference in sound when you fire your DSP Plutos off of the front wall (as opposed to facing the 2"'s towards you)? In this case the < 3-4 kHz info is arriving earlier than the rest of the treble energy, similar to the Watson setup. You've mentioned before that the image formed is convincing. I've felt the same with 3" drivers. In fact it seems that I much prefer to redirect their treble.
It's interesting. If the brain would like to analyse harmonics first and they reveal the tweeter locations, depriving them of this information makes a lot of sense.
I might try an experiment where I play loud speakers at the same time as headphones, with a delay on the headphones. Headphones have little attenuation before 1 kHz. I'll be curious to see if the localization cues from the speakers persist and the treble energy from the headphones through the HRTF zone and beyond just corrects timbre.
It's interesting. If the brain would like to analyse harmonics first and they reveal the tweeter locations, depriving them of this information makes a lot of sense.
I might try an experiment where I play loud speakers at the same time as headphones, with a delay on the headphones. Headphones have little attenuation before 1 kHz. I'll be curious to see if the localization cues from the speakers persist and the treble energy from the headphones through the HRTF zone and beyond just corrects timbre.
Please try it yourself. Play the same sound through a second speaker and raise its level. Try different delays, 5ms and higher. Precedence isn't as rock solid as literature might suggest.
Interesting observation.
How does spatial reproduction change when placing the speakers as close as possible to the front wall?
Of course EQ changes somewhat, but is easy to adjust.
Well behaved tweeters in stereo need only give themselves away when signals to tweeters are highly uncorrelated.
When tweeters are fired at reflective wall, speakers become virtual image from behind wall, with narrower apparent aperture.
Headphone delay should make for nice tricks.
Regards,
Andrew
Well behaved tweeters in stereo need only give themselves away when signals to tweeters are highly uncorrelated.
When tweeters are fired at reflective wall, speakers become virtual image from behind wall, with narrower apparent aperture.
Headphone delay should make for nice tricks.
Regards,
Andrew
In case of Pluto type speaker, bulk of information cues are firing from tweeter, and dominant effect is EQ of low end. Pushed right up to wall imaging remains very good, and pushed up to wall with tweeter pointed at wall works virtually the same.
Asymmetrical set up of each speaker relative to walls and corners has greatest impact on imaging.
Regards,
Andrew
A possibly interesting setup with, e.g. Pluto, would be to align the listening axis with the corner of a room, place speakers in line perpendicular to the walls, and then place them facing to or away from the wall. It could increase the channel separation a little since L and R ears will get first dibs on the reflections from the respective L and R speakers.
Do you perceive a difference in (phantom) source distance or stage depth with the speakers at the front wall vs. speakers farther away from front wall?
I love that we're talking about Griesinger's research. He takes audio engineering to a higher level. I've been obsessed about imaging for decades, so he's one of my favorites.
On the issue of the stereo triangle not being Hi-Fi, well, I think we would all agree that perfection in reproduction isn't an option. Yet we can create an awfully likable reproduction anyway. If you're near the sweetspot, it should only be the frequencies who's wavelengths are short that will likely be a problem, above say 800HZ. I've read and believe that above about 6kHZ our ability to localize a sound source falls off rapidly, although the 8kHZ region is apparently used as a cue for height information. There should be an equation where we can dump this info in, along with the knowledge that we interpret image location by timing comparisons below about 800HZ, and by amplitude comparisons above about 1.5kHZ.
What does this add up to? Well, I think we need to have interaural cancellation to clean up and make effective good imaging in the lower midrange, and perhaps limited dispersion to minimize room reflections in the uppermidrange, and perhaps omni-directionality above 6kHZ.
I'm building a speaker right now that has two crossover points; 150HZ (4 pole) and 7kHZ (1 pole). Since it's a "soundbar" cabinet that goes under my TV set, I'm opting for the Polk method of inter-aural crosstalk reduction. I like the Polk method because it's not as critical exactly where your head is.
I like using inter-aural cancellation in a "soundbar" since the soundbar also has an L+R driver in the middle that should help clean up the center image, which the Carver (electronic) version of inter-aural cancellation left sounding a bit "phasey". Although I'm also using an L-XR circuit (matrix - which cancels most of the L+R info out of the non-L+R drivers) to give better separation between the center driver signal and the rest of the drivers (5 total across the front of the cabinet handling 150HZ - 7kHZ), I may find that it's not perfect (some interference patterns), when all the signals add up. But I'm excited about it anyway. These Peerless/Vifa drivers that Roger Russel is using appear to be excellent for going 150-7kHZ. No crossover in that whole range. and they get a bit directional in the uppermidrange due to their physical size (3 inch), so less room reflections (maybe a good thing?) in the uppermidrange. The highest quality recordings should be less damaged. Cabinet diffraction and room acoustics will surely screw it up a bit, but it may be pretty good anyway. I always glue absorbtive material to the outside front surface of my cabinets to minimize diffraction effects. I'll report back when it's up and running. It will be a while.
On the issue of the stereo triangle not being Hi-Fi, well, I think we would all agree that perfection in reproduction isn't an option. Yet we can create an awfully likable reproduction anyway. If you're near the sweetspot, it should only be the frequencies who's wavelengths are short that will likely be a problem, above say 800HZ. I've read and believe that above about 6kHZ our ability to localize a sound source falls off rapidly, although the 8kHZ region is apparently used as a cue for height information. There should be an equation where we can dump this info in, along with the knowledge that we interpret image location by timing comparisons below about 800HZ, and by amplitude comparisons above about 1.5kHZ.
What does this add up to? Well, I think we need to have interaural cancellation to clean up and make effective good imaging in the lower midrange, and perhaps limited dispersion to minimize room reflections in the uppermidrange, and perhaps omni-directionality above 6kHZ.
I'm building a speaker right now that has two crossover points; 150HZ (4 pole) and 7kHZ (1 pole). Since it's a "soundbar" cabinet that goes under my TV set, I'm opting for the Polk method of inter-aural crosstalk reduction. I like the Polk method because it's not as critical exactly where your head is.
I like using inter-aural cancellation in a "soundbar" since the soundbar also has an L+R driver in the middle that should help clean up the center image, which the Carver (electronic) version of inter-aural cancellation left sounding a bit "phasey". Although I'm also using an L-XR circuit (matrix - which cancels most of the L+R info out of the non-L+R drivers) to give better separation between the center driver signal and the rest of the drivers (5 total across the front of the cabinet handling 150HZ - 7kHZ), I may find that it's not perfect (some interference patterns), when all the signals add up. But I'm excited about it anyway. These Peerless/Vifa drivers that Roger Russel is using appear to be excellent for going 150-7kHZ. No crossover in that whole range. and they get a bit directional in the uppermidrange due to their physical size (3 inch), so less room reflections (maybe a good thing?) in the uppermidrange. The highest quality recordings should be less damaged. Cabinet diffraction and room acoustics will surely screw it up a bit, but it may be pretty good anyway. I always glue absorbtive material to the outside front surface of my cabinets to minimize diffraction effects. I'll report back when it's up and running. It will be a while.
Last edited:
Sure. That's why I invited everyone to imagine what happends to the harmonic phases in the stereo interference field
Of course they are no longer in coherence. Thus stereo cannot be considered as Hi-Fi Or, if stereo is Hi-Fi, then Hi-Fi is something else. It is not a preservation of the original sound.
- Elias
At first, be sure you have understood Blumlein's stereo patent GB394325 from 1931.
In the ITD (interaural time difference) range (which you refer as below 800Hz) stereo is different from natural hearing in the way the timing cues are generated. At low freqs stereo relies on amplitude panning of the channels. The timing cues are generated on the spot at the listening position of the stereo triangle. It's basic physics, summing of two wavefields, nothing complex.
This leads to the objection of the need of cross talk cancellation in the ITD range. Because, interestingly, ITD range is the only freq range stereo does work as intended !!
I think cross talk is killing stereo at higher freqs, in the ILD range, above about 2kHz. So the effort should be placed there.
- Elias
Last edited:
I did read the original Blumein patent, and many papers on this subject from many other authors over the years. In the upper midrange amplitude comparisons are the main way we determine image location. In the lower midrange, timing comparisons are the main cue for image location. Am I missing something here?
It's highly unlikely that "stereo" will ever be perfect, but mine is definitely "Hi-Fi". And who cares what the mixing engineer thought was optimal. If you can make it sound even better, by adjusting tone controls, using inter-aural cancellation, or surround sound, then don't be fool.Sure. That's why I invited everyone to imagine what happends to the harmonic phases in the stereo interference field
Or, if stereo is Hi-Fi, then Hi-Fi is something else. It is not a preservation of the original sound.
- Elias
Propably not much ! Just a notion that Blumlein recording does not contain low freq timing cues. It has supposedly been recorded using a XY microphone.
Recordings with low freq timing cues are not compatible with stereo triangle reproduction.
Nevertheless, stereo triangle cannot reproduce "The phase coherence of harmonics in the vocal formant range, ~630Hz to 4000Hz". So sad
Still these cues form a fundamental set of perception information according to Griesinger.Having read the powerpoint presenation - it reads like two totally separate speeches.
The first 50 pages concern the title presentation, specifically the study of listener perception with respect to acoustics (or NOT stereo reproduction).
The last few pages (51-64), are just about stereo reproduction and there doesn't seem to be any real "tie-in" with the title presentation.
The first 50 pages concern the title presentation, specifically the study of listener perception with respect to acoustics (or NOT stereo reproduction).
The last few pages (51-64), are just about stereo reproduction and there doesn't seem to be any real "tie-in" with the title presentation.
It's a *lack* of cross-talk at higher freq.s.. and to much cross-talk below that upper freq. bandwidth.
Of course if the recording is binuaral (or a synthetic version), then there is to much cross-talk at any freq.. though particularly so at freq.s below 3 kHz.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.