A couple of comments. The smoothness of decay is related to the bandwidth of the measurement. I think these are broad band decay curves so they will be a little smoother because of that. Still, I've done my share of RT measurements and they tend to be smooth for upper frequencies in any room size I've measured. (Slope may vary with decay)
Again I'd say that room size is irrelevant except for the encroachment of the Schroeder frequency. Architectural acoustics is scalable. a 1/10th scale or 1/100th scale model will measure much the same as a full size room, when the wavelengths are scaled appropriately. (Yes, HF air absorption needs to be considered.) All rooms are diffuse above a certain frequency!
I've measured RT of my home room and multiple curves from different positions and with speakers aimed in different dirction give the same result from about 160 Hz.
I've worked (briefly) in architectural acoustics. Many factors other than RT have become useful in the last several decades, but RT is still one of the important ones. RT is just the broadest look at liveness. Other newly popular factors such as G (loudness of a given sound at a given distance into the hall) cann't be separated from RT. Read Beranek on concert hall design and you will find that RT and RT variation with frequency are still primary measurements and part of the objective ranking of concert halls. Old but not obsolete!
There is an ideal RT for most listening rooms. It is in the 3. to .5 range depending on size. You might be better considering mean alpha because then you can factor out room volume. A mean alpha of about .4 (40% of energy absorbed per "bounce") is a good target and works over a very broad range of sizes, from domestic room up to large movie theaters!
Rooms with little damping have an annoying, echoey, ringing quality. They are difficult for conversation and uncomfortable in general. I'm not sure why you recommend making a room as lively as possible at HF? I'll bet I can separate the good ones from the bads ones with an RT measurement.
David S.
Last edited:
Member
Joined 2009
Thank you for the explanation, it's much clear now. I did not understand that the "stereo" signals are totally different.
They actually look different on Winamp's Spectrum Analyzer but I don't want to pursue the issue, I think it's pointless to this discussion.
You're right, the first two ETC images are indeed broadband decay. I can also provide narrow band ETC's if anyone is interested.
A cursory look at them (which I hadn't done until now) shows 2Khz, 4Khz, and 8Khz more or less perfectly straight decay from about -5dB to -50dB, 500Hz and 1Khz is slightly bumpy especially around 250ms but still approximates a straight line, 250Hz and lower has quite serious periodic humps in the decay. Whether this is comb filtering or room modes, or just an artefact of the way ARTA processes the signal I don't know.
Usefulness of an RT60 measurement in a small room at or below 250Hz is questionable though, I only take notice of it from about 500Hz up, and certainly not below 250Hz.
Last edited:
So, up to date statistics on the high frequency pink noise stereo phantom localisation tests:
Perceived high freq image at the center only:
DBMandrake
erjee
test subject by erjee
markus76
dantheman
boris81
Paul W
Humdinger
----
Perceived high freq image at the tweeters:
Radugazon
1st test subject by radugazon
2nd test subject by radugazon
3rd test subject by radugazon
Elias
----
Borderline cases, perceived high freq images at the tweeters and center:
Rudolf
----
N = 14, out of which 43 % perceived high freq images at the tweeters. Is this insignificant ?
- Elias
Perceived high freq image at the center only:
DBMandrake
erjee
test subject by erjee
markus76
dantheman
boris81
Paul W
Humdinger
----
Perceived high freq image at the tweeters:
Radugazon
1st test subject by radugazon
2nd test subject by radugazon
3rd test subject by radugazon
Elias
----
Borderline cases, perceived high freq images at the tweeters and center:
Rudolf
----
N = 14, out of which 43 % perceived high freq images at the tweeters. Is this insignificant ?
- Elias
I thought this hi-passed pink noise test was interesting and revealing. I had read somewhere that above about 5-6kHZ, our ear-brain mechanism loses the ability to pinpoint where an acoustic image is coming from. It initially left me wondering if tweeters above that frequency needed to be wide dispersion. I think that although we may not have pinpoint image recognition in that frequency range, we are very sensitive to it being there and being diffuse.
Somebody asked what is meant by "Flanker Subs".
I love this flanker concept because it makes sense and works real well in my situation. In the above "Room Modes..." link, it's explained very well. I especially liked the part about putting flankers that handle the 100HZ - 200HZ at a different elevation, since ceiling bounce is likely to be a main culprit, due to the size of the wavelength involved.
This led me to appreciate even further that my lower mid drivers handling 100HZ - 1.4kHZ are in the form of four 5 inch drivers in a vertical array approx 3 ft. high (and tilted back by 10 degrees to minimize floor bounce and timing variations). I did that for several other reasons before realizing that it helps with this issue as well. I felt that I must be kidding myself somehow when I measured such a flat response at the listening position from 20HZ to 20kHZ... No major problem in the lower mid or bass. What are the chances? Now it makes a little more sense. VERTICAL ARRAYS ARE BACK, BABY!!!
This is not at all my experince. I mean if you are talking about a bathroom perhaps, but any real "living" room is going to have a lot of natural damping above 500 Hz, even from the listeners, etc. so an "echoey, ringing quality" IS NOT how I would describe it. I would say that because larger rooms have inherently longer reverb times, a lively small room sounds much bigger than it really is. I find dead small rooms annoying and claustrophobic for just this reason. Like headphones when the signal is off, or an anechoic chamber.
Its also about where the damping is located as well. I put a massive amount of damping behind the speakers because I find any sound coming from that direction, other than the direct sound from the speakers, interfeers with the imaging. (This damping alone would probably bring down the overall RT60, but its also going to yield a distinctly non-difuse situation.) Aft(er) of the speakers I use no damping at all (well a floor bounce absorber) because what is naturally there is enough. Thus there is a fairly long reverb time of sound coming from the sides and back and this does give the impression of the spaciousness of a much larger room. But as I said before, you CANNOT put most speakers in this room because it will just amplify all of their problems.
Its all about the loudspeaker/room design and not, the room, and then the loudspeaker.
I suppose T60 might have some uses as is being suggested, its just not a measure that I see much use for in a small room. But to think that you can "separate the good ones from the bads ones with an RT measurement" - a single number independent of the loudspeakers being used - is simply absurd IMO. That's like saying that the early reflection pattern doesn't matter, a claim that you sometimes seem to believe at times. I simply don't buy it. Maybe in some situations, like classical music over omni-directional speakers, but its not going to be true for all recordings, loudspeakers and room designs.
I have found in general that people who listen to large orchestral works do look for very different things in their playbacks than I look for. Only a very small fraction of my customers fall into that camp however, and for the most part almost no one who has come to audition has brought this kind of music. I find that my approach is no worse for large orchestral works, but it doesn't shine either. There is far too much "space" in the recordings for my room and its design to be appreciable. But on other types of studio music where there is not a strong intent to highlight the recording "space" my approach works exceptionally well.
And to Simons question, if I have a room that is coupled to another room, as is very common in homes, then the decay curve will not be a straight line, even broadband. If you have the luxury of being able to close off the space from any other spaces then thats going to work a lot better.
I've been in several rooms that ringed so bad that when you clapped your hands, it sounded like glass was shattering above your head and all around you. And you're right, they sound bad even just in conversation. Directional speakers helped, especially with the ceiling slap. This is probably because the directional speakers I'm talking about have fairly narrow vertical patterns. But even then, it was still important to get some stuff in the room to damp it as much as possible. Rugs, curtains and lots of furniture with big fluffy cushions.
Do you know Rick Talaske, from Chicago? I don't know if he is still practicing, but he wouldn't even discuss T60. It was completely unimportant to him, and this was at least a decade ago. Another guy I know who worked in NY, David Kahn, was the same way. They wanted nothing to do with T60. Maybe they have changed, I haven't seen either one in a very long time.
No, it's not but it's also not clear what causes the difference in perception. The room, the speaker, differences in perception or a combination.
By the way, with toe-out I also had the borderline case like Rudolf.
P.S. Did you listen to the sample over headphones? Do you also perceive two different sources at each ear, a sharp phantom image in the middle of the head or something between those two extremes?
RT60/RT30 is interesting. It certainly worked for me.
Here you can see the difference in my listening room of wall treatments. In the 1st post you'll see the simulations from CARA software. They pretty much agree with the before and after measurements (much to my delight).
Reducing the T30 times in the midrange had a huge audible effect on music and speech in the room. It went from the bare room "ping" to a fairly comfortable space. Still room for improvement, but a big step in the right direction. The better measurement sounds better, just as you might expect.
Here you can see the difference in my listening room of wall treatments. In the 1st post you'll see the simulations from CARA software. They pretty much agree with the before and after measurements (much to my delight).
Reducing the T30 times in the midrange had a huge audible effect on music and speech in the room. It went from the bare room "ping" to a fairly comfortable space. Still room for improvement, but a big step in the right direction. The better measurement sounds better, just as you might expect.
Somebody asked what is meant by "Flanker Subs".
I love this flanker concept because it makes sense and works real well in my situation. In the above "Room Modes..." link, it's explained very well. I especially liked the part about putting flankers that handle the 100HZ - 200HZ at a different elevation, since ceiling bounce is likely to be a main culprit, due to the size of the wavelength involved.
This led me to appreciate even further that my lower mid drivers handling 100HZ - 1.4kHZ are in the form of four 5 inch drivers in a vertical array approx 3 ft. high (and tilted back by 10 degrees to minimize floor bounce and timing variations). I did that for several other reasons before realizing that it helps with this issue as well. I felt that I must be kidding myself somehow when I measured such a flat response at the listening position from 20HZ to 20kHZ... No major problem in the lower mid or bass. What are the chances? Now it makes a little more sense. VERTICAL ARRAYS ARE BACK, BABY!!!
The ability of vertical arrays to mitigate floor/ceiling bounce and higher frequency (100Hz to 200Hz) room modes is partly what drove me to develop and recommend the flanking sub configuration. Back when I was first studying Welti's work on multiple subwoofers, and during some discussions with Geddes about his pseudo-random approach, I happened to be asked to measure a line array for a friend. I noticed it was very smooth in the lower midrange, where every other point source speaker on a stand had ripple from self-interference. I quickly realized that a truncated vertical array (even just two drivers) goes a long way towards mitigating floor bounce. This problem sort of goes hand-in-hand with the room mode problem, and the solutions are somewhat similar too.
My constant directivity cornerhorns are also very smooth for the same reason vertical arrays are - the woofer and midhorn are blended in the lower midrange region, and they are separated a couple feet in the vertical. This smoothes vertical modes and self-interference from the floor or ceiling in the 100Hz-300Hz range. They are also nestled back in the corner so there is no self-interference from the walls extending from the apex of the corner.
The combination of these two principles brought me to the flanking subs approach. It smoothes the lower midrange notch that is nearly always present from a point source loudspeaker on a stand. The self-interference from the wall behind the speakers, the floor bounce notch and the vertical modes all conspire to produce ripples in the lower midrange. They could be mitigated like the array, but by offsetting a little bit in the horizontal as well as the vertical - displacement in all three dimensions - the ripple caused by self-interference from the walls could be smoothed in addition to the floor bounce notch.
I agree with you that directional speakers can work better in lively rooms than less directional speakers. And I agree that reflections from the wall behind the speakers are unwanted. So I suppose using absorbent material behind the speakers is a good thing. It certainly would be for an omnidirectional speaker.
But I question the usefulness of absorbent material behind a directional speaker. The high frequencies are already attenuated greatly behind the speaker by its directivity. The lower frequencies aren't damped well by the stuff behind it, unless it's huge. Are you saying you put a mound of pillows or something several feet thick on the wall behind the speakers? Something thick enough to damp the lower midrange where your speakers are radiating strongly in all directions?
You feel reducing T30 improves things but at the same time argue against Earl's RFZ approach?
That will make the statistic 50 / 50 %
It is a toss of a coin if the stereo will work at high freqs or not !
It totally depends on the headphones. I don't have in-ear phones at hand, but these whimpy pads on top of my ear flaps make the high freq hiss to come outside my head ! I guess there is signal leakage into the pinna. The low freq image is inside my head.
- Elias
I think that we need to get on the same page here. Your data shows a RT time of almost 1 sec. That is enormous (Almost unbelieveable for a "small" room) and either the room is very large or its a bomb shelter. I have been talking about rooms that are on the "lively" side of reality not rooms that are off the chart like the one that you showed.
So in those exampples, the loudspeakers are identical, the room hasn't changed and so if the early reflection pattern does not matter, then the two situations should sound that same. Right?
So do they sound the same?
If they don't then there is more to the problem than just the rooms RT60 and or the loudspeakers polar reponse, since neither of those things changed.
And if you think that's bad, you should have heard it before the heavy carpet and pad!
0.85 seconds at 1.6Khz was the peak. Yes it sounded awful, but not different than you might expect in a room with 4 bare parallel walls and a plywood ceiling at 11'. Do you see it as unusual for a bare room? No furniture expect the speakers.
EDIT: The sim showed the peak an octave or more low, but the times were right. Bad sim, bad measurement or something else?
0.85 seconds at 1.6Khz was the peak. Yes it sounded awful, but not different than you might expect in a room with 4 bare parallel walls and a plywood ceiling at 11'. Do you see it as unusual for a bare room? No furniture expect the speakers.
EDIT: The sim showed the peak an octave or more low, but the times were right. Bad sim, bad measurement or something else?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.