• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Dr. Toole's comments in anticipation to his new book on Acoustics and Psychoacoustics

fas42

Major Contributor
Joined
Mar 21, 2016
Messages
2,818
Likes
191
Location
Australia
Room Reverberation Experiment:

Mono signal applied to both (stereo) speakers, recorded waveform.

4 ms of 1000Hz and then 4 ms of 3000Hz, then about 80 ms of room contribution only.

Plots: Signal, JBL LSR 308, MartinLogan reQuest (dipole)
Ray, where was the mic positioned for these?
 

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,204
Likes
16,985
Location
Riverview FL
And if you were to keep a signal going for a second or two in order to let the reverb 'build up', and then stopped the signal..?

Here's 10, 50, 100, 200 and 500 ms tones, aligned on the release, with decibel presentation of the decay.

upload_2017-6-20_1-10-58.png


100, 200, 500 ms look about the same

50 ms looks a little odd, maybe some external noise (didn't hear it).

10 ms looks like it is short enough to have gaps in the echoing.

I didn't go longer because they all seem to have about the same decay length already.

---

300 ms of decay only (waveform) amplified by 19.25dB for visibility:

upload_2017-6-20_1-48-25.png


I conclude there isn't much "buildup" of sound, beyond a single fillup.

My room seems to "fill" at some point between 50 and 100 ms of steady tone, looking at the amplitudes and decay patterns.

Room is about 18 x 14 x 9 and the left rear corner is open to the kitchen and other areas.
 
Last edited:

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,204
Likes
16,985
Location
Riverview FL
Ray, where was the mic positioned for these?

10 feet 2 inches from each speaker, on the top of the couch, pointing at the phantom center, 7" of rockwool behind, about 5 feet from the wall behind, approximating my critical listening and dozing off position, and accurately acoustically centered to less than .02 ms (6.8 mm), which seems to be the measurement limit of REW, and the increment available for channel delay on the miniDSP.

upload_2017-6-20_1-22-18.png


Oh, now I see.
.02 ms is one (rounded off) sample time at 48kHz :
1 / 48000 = 0.00002083333333 seconds
 
Last edited:

Cosmik

Major Contributor
Joined
Apr 24, 2016
Messages
3,075
Likes
2,180
Location
UK
Here's 10, 50, 100, 200 and 500 ms tones, aligned on the release, with decibel presentation of the decay.

View attachment 7580

100, 200, 500 ms look about the same

50 ms looks a little odd, maybe some external noise (didn't hear it).

10 ms looks like it is short enough to have gaps in the echoing.

I didn't go longer because they all seem to have about the same decay length already.

---

300 ms of decay only (waveform) amplified by 19.25dB for visibility:

View attachment 7582

I conclude there isn't much "buildup" of sound, beyond a single fillup.

My room seems to "fill" at some point between 50 and 100 ms of steady tone, looking at the amplitudes and decay patterns.

Room is about 18 x 14 x 9 and the left rear corner is open to the kitchen and other areas.
When you say "tones" do you mean the two frequencies you mentioned earlier? Could you repeat the experiment with full spectrum noise?
(thanks!)
 

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,204
Likes
16,985
Location
Riverview FL

Cosmik

Major Contributor
Joined
Apr 24, 2016
Messages
3,075
Likes
2,180
Location
UK
No, actually this was 2kHz.



Yes. Can do. It wobbles more afterwards.
I was thinking noise might stimulate more modes or whatever.
 

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,204
Likes
16,985
Location
Riverview FL
I made 1 second of 20-20k pink.

It seems to have "carried over" (unexpectedly?) for 10ms, but then decayed as well as the tones above. Decibel values dropped at essentially the same rate.

Tone from before on top, noise on bottom:

upload_2017-6-20_5-1-58.png


Decibel:

upload_2017-6-20_5-7-38.png
 

Cosmik

Major Contributor
Joined
Apr 24, 2016
Messages
3,075
Likes
2,180
Location
UK
I made 1 second of 20-20k pink.

It seems to have "carried over" (unexpectedly?) for 10ms, but then decayed as well as the tones above. Decibel values dropped at essentially the same rate.

Tone from before on top, noise on bottom:

View attachment 7583

Decibel:

View attachment 7584
Or is the "carry over" actually reverberation? (We were perhaps expecting it to be as loud, or louder, than the signal itself..? - according to Linkwitz)
 

The Smokester

Active Member
Joined
May 11, 2017
Messages
136
Likes
39
Location
SF Bay
I think the difficulty is that the numbers in your A(f)=k(f) a(f) are complex.
As shown above, the direct sound can still obey A(f)=k a(f), but integrated over some longer time interval you measure k(f)

Yes. Definitely complex numbers are allowed.

To what difficulty are you referring? And over what time interval do you suggest be measured? And why?
 

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,204
Likes
16,985
Location
Riverview FL
Or is the "carry over" actually reverberation? (We were perhaps expecting it to be as loud, or louder, than the signal itself..? - according to Linkwitz)

I don't know.

I tried all 88 fundamentals of a piano (at once) for one second. No holdover like seen with the pink. Not as clean as with the single tone.

upload_2017-6-21_1-49-37.png


upload_2017-6-21_1-50-55.png
 

Cosmik

Major Contributor
Joined
Apr 24, 2016
Messages
3,075
Likes
2,180
Location
UK
I don't know.

I tried all 88 fundamentals of a piano (at once) for one second. No holdover like seen with the pink. Not as clean as with the single tone.

View attachment 7596

View attachment 7597
But the upshot is that the reverberated stuff is roughly the same level as the direct in a typical room. What does that mean?

In my way of looking at it, if the speaker was 'ideal' (a small box with whatever dispersion angle we choose, covering all frequencies the same), it would mean very little. The speaker would sound correct in any room's acoustics, and it would be a fool's errand to think we could EQ it to sound better. This would be just the same as if we got a singer in for a live performance: we wouldn't ask them to wear a cotton mask "just to shave a bit off the top end" depending on the acoustics.

But speakers aren't ideal: the dispersion isn't uniform with frequency. At the same time, we "hear through the room" (Dr. Toole). What I think this gives us is a conundrum that cannot be solved by an in-room frequency response target curve that implicitly treats all speakers with differing dispersion patterns the same. Nor is it solved by a speaker that is EQ'ed to be flat in an anechoic chamber. It is something else, and whatever it is, it will never be technically 'correct'. The way I see it, the optimum is likely to be 'Whatever you do to EQ the speaker to be flat in an anechoic chamber, but a bit less'. How much less? I think you can only set that depth by ear in the listening room - which is what I do for my system, and it isn't too difficult to find the right setting.
 
Last edited:

The Smokester

Active Member
Joined
May 11, 2017
Messages
136
Likes
39
Location
SF Bay
From Linkeitz:
"Hearing has evolved as an effective survival
mechanism for detecting direction, distance, and
meaning of a source of sound in an environment where
there may be multiple sources of sound and reflections
[1]. Hearing still operates like that today when the head
automatically turns towards a surprising or unfamiliar
sound to figure out what is causing the sound and how
far away it is. Having identified the source and its
proximity the response is either to run, watch out,
ignore it or be annoyed."

A driving feature of his using dipoles is that the radiation pattern is directional. So the brain can more easily separate the direct sound from the diffuse reflections of the room.

He makes the point that headphones are weird and unnatural because when the listener turns his/her head there is no change in sound.

It is well accepted that an over-damped room is unnatural.and requires growing accustomed to (at the very least).
 

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,204
Likes
16,985
Location
Riverview FL
A driving feature of his using dipoles is that the radiation pattern is directional.

I (think) I measure a 10-15dB difference in the ratio of direct/reflected between the dipoles and monopoles in my (untreated sidewall) room.
 

Fitzcaraldo215

Major Contributor
Joined
Mar 4, 2016
Messages
1,440
Likes
632
What I think this gives us is a conundrum that cannot be solved by an in-room frequency response target curve that implicitly treats all speakers with differing dispersion patterns the same. Nor is it solved by a speaker that is EQ'ed to be flat in an anechoic chamber. It is something else, and whatever it is, it will never be technically 'correct'. The way I see it, the optimum is likely to be 'Whatever you do to EQ the speaker to be flat in an anechoic chamber, but a bit less'. How much less? I think you can only set that depth by ear in the listening room - which is what I do for my system, and it isn't too difficult to find the right setting.

I don't think anyone disagrees that the EQ target curve above transition frequency ideally should be adjusted by ear. I think Toole, Olive and Amir say so. And, so, apparently, do you, too. Not all Room EQ tools support target curve mods, but the better ones that are purchased separately from AVRs, etc. usually all do. However, even where you can modify the target curve to taste, it makes a lot more sense to use something similar to the downward sloping B&K curve as a default starting point rather than a flat target when beginning the by ear adjustment process. Flat is just wrong sounding, except in true anechoic or partially anechoic situations like near field listening.

The degree of slope or other issues might need to be tweaked in individual cases based on speaker dispersion, room characteristics, listener taste, etc. But, broad listener preference for some form of the downward sloping curve seems quite well established.

However, it happens that I have been quite happy with the default downward sloping Dirac curve with my ML dipole hybrids. A friend also with Dirac and Revel Salon2s has spent a huge amount of time tweaking his target. Personally, I think his system sounds best with the default curve rather than customized, but it is his system and he is still tweaking by ear.
 

The Smokester

Active Member
Joined
May 11, 2017
Messages
136
Likes
39
Location
SF Bay
I once went to a piano concert in the oldest church in Paris, Abbey de Saint-Germain des Prés. When the pianist start to play, I could not hear it as music because the many delayed reverberations from the ?stone? walls overwhelmed the direct sound from the piano. By the end of the concert I heard only the music.

It was a beautiful concert, a wonderful memory and a vivid illustration to me of the human mind's ability to adapt.
 

Brad

Active Member
Joined
Nov 8, 2016
Messages
114
Likes
35
Yes our brains are able to generate filters, but it takes a little time.
The problem with complex numbers, is that they are more difficult to interpret.
To resolve the issue of whether the response A(f)=k(f) a(f) is a linear system or not, (as the trivial answer is clearly that it's non-linear), if you were able to capture only the direct response (ie anechoic response) you would get A(f) = k a(f). The reflections appearing after the direct response add linearly, but since they are coherent, non-linear terms are generated. So it's a linear coherent response.
 

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,204
Likes
16,985
Location
Riverview FL
The reflections appearing after the direct response add linearly, but since they are coherent, non-linear terms are generated. So it's a linear coherent response.

I would not consider them coherent.

(I can be wrong)
 
Last edited:

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,835
Likes
16,497
Location
Monument, CO

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,204
Likes
16,985
Location
Riverview FL
Last edited:

Brad

Active Member
Joined
Nov 8, 2016
Messages
114
Likes
35
Coherent in the signals-and-systems sense, though they are not completely coherent if they don't have the same phase. See e.g. https://en.wikipedia.org/wiki/Coherence_(signal_processing)

I do not follow how linear signals added linearly produce a nonlinear result, however.

You are right, it was a bit early in the morning. Coherent signals produce a 2Acos(theta)^0.5 term. I was thinking of the sin(theta)cos(theta) term, which is 2sin(2theta), but that needs a nonlinearity

Anyway this is the coherence I was thinking about
https://en.m.wikipedia.org/wiki/Coherence_(physics)#Temporal_coherence

The direct signal and a reflected signal will have temporal coherence as there is a fixed phase relationship
 
Last edited:
Top Bottom