Paul Hales on room equalization on Soundstage! Network

[kemmler3D]

Please explain how a room mode affects a sound that has not interacted with the room. Or are you proposing that the sound “knows” the walls are there even though it hasn’t had time to reach them?

That objection isn't wrong afaik but focuses on the wrong thing. The larger point is it's hard to perceive the direct sound separately from the room at low frequencies, and so there's arguably little to nothing to save from the EQ in perceptual terms, if that makes sense.

 

[Keith_W]

If the room has a mode that causes a peak at 60 Hz and you implement a notch filter at 60 Hz to compensate, the direct sound arrives with a big notch at 60 Hz, not with flat response. He claims that one result is a significant loss of impact in the bass.

Are you sure he said that? That seems to be a very strange thing for a competent speaker designer to say. Let's say you were in a free-field (meaning sound goes past you, never to return) so ALL you hear is the direct sound. It arrives with a big notch at 60Hz. You are less likely to hear it, since your hearing is less sensitive to dips, especially if those dips are narrow in Q. But this is in a room, your brain will perceptually fuse the direct sound with the reflections.

I didn't listen to the whole thing, only half of it. I mostly agree with what I heard, especially when he clarified that he isn't anti-DSP, he is anti-"DSP that is applied to poorly taken measurements".

 

[NTK]

Here are my (1-D) simulations of excitation of an axial room mode. The excitation signals are tone bursts of 1 to 6 cycles. Note that these tone bursts are ramped up and down using a raised cosine window (Hann window) to reduce the abruptness of the start and stop transitions to reduce the high frequency contents. These simulations are largely based on the parameters I used in my visualizing room modes thread.

For each animation of the simulation results, the top plot shows the pressure response in the room. The bottom plot shows the sound pressure vs time plot for 3 locations. With 1 cycle, the excitation is too short to activate the room mode. We basically only have reverberations. When far enough away from the walls, by the time the reflection wave comes back, the forward wave is no longer active, so there is no constructive or destructive interference happening. However, at the green location (a null position), it is close enough to the right wall that some destructive interference can be observed. Due to the front wall reflection (source is 0.5 m from the front wall), the spatial waveform of the sound pressure is a "spread-out" version of a sine wave. These simulations showed, in this case, it takes 5-6 cycles to start to fully activate the room mode. The activation of the room mode is easiest to see from the green location (null) response.

I'll leave readers to make their conclusions.

1 cycle:

2 cycles:

3 cycles:

4 cycles:

5 cycles:

6 cycles:

 

[gnarly]

I think what's largely missing from this discussion, is the duration of transient impact from explosive bass.
In the Hales interview, starting at about 34 min in he talks about this, saying 90% of a kick drums energy happens in 20ms.

A kick drum hit, or a big bass drop, are short term events, more full-range than not.
To the extent room-correction of modal behavior alters the steady state anechoic performance of the speaker,
it also alters the short term transient behavior of the speaker, the 20ms response, before modal behavior begins.
Transient gets sacrificed for steady state, imo/ime.

 

[NTK]

Transient gets sacrificed for steady state, imo/ime.

That's the nature of the beast of listening to reproduced sound in a reverberant room.

In my example of simulation above, it took roughly 6 cycles of excitation 64 Hz to activate an axial room mode in a 8 m long room. The duration is 6/64 = 0.094 s, and in that time sound travels 0.094×343 = 32 m, which is about 4 room lengths. So roughly 4 full room bounces.

At a tempo of 120 bpm, a quarter note is 0.5 s, so 0.094 s is a little less than a sixteenth note. Therefore, we may reason that any music notes of 1/16 or longer may be affected by the room mode.

So "transient" vs boomy notes. You decide.

 

[kemmler3D]

Here are my (1-D) simulations of excitation of an axial room mode. The excitation signals are tone bursts of 1 to 6 cycles. Note that these tone bursts are ramped up and down using a raised cosine window (Hann window) to reduce the abruptness of the start and stop transitions to reduce the high frequency contents. These simulations are largely based on the parameters I used in my visualizing room modes thread.

For each animation of the simulation results, the top plot shows the pressure response in the room. The bottom plot shows the sound pressure vs time plot for 3 locations. With 1 cycle, the excitation is too short to activate the room mode. We basically only have reverberations. When far enough away from the walls, by the time the reflection wave comes back, the forward wave is no longer active, so there is no constructive or destructive interference happening. However, at the green location (a null position), it is close enough to the right wall that some destructive interference can be observed. Due to the front wall reflection (source is 0.5 m from the front wall), the spatial waveform of the sound pressure is a "spread-out" version of a sine wave. These simulations showed, in this case, it takes 5-6 cycles to start to fully activate the room mode. The activation of the room mode is easiest to see from the green location (null) response.

I'll leave readers to make their conclusions.

1 cycle:
View attachment 548032
2 cycles:
View attachment 548033
3 cycles:
View attachment 548034
4 cycles:
View attachment 548035
5 cycles:
View attachment 548036
6 cycles:
View attachment 548037

Thanks, super helpful and I think seeing this makes the argument a lot more intuitive, and I think establishes that the objection to EQ is at least plausible.

I think what's largely missing from this discussion, is the duration of transient impact from explosive bass.
In the Hales interview, starting at about 34 min in he talks about this, saying 90% of a kick drums energy happens in 20ms.

A kick drum hit, or a big bass drop, are short term events, more full-range than not.
To the extent room-correction of modal behavior alters the steady state anechoic performance of the speaker,
it also alters the short term transient behavior of the speaker, the 20ms response, before modal behavior begins.
Transient gets sacrificed for steady state, imo/ime.

20ms is (as a rule of thumb) the shortest sound where it's perceptible as tonal, i.e. not just a click. Bass notes are pretty much always much longer than this... load up a sample in Audacity or something and trim it to 20ms, you'll see what I mean.

However, I do think the idea of losing impact from the room EQ is pretty plausible at this point, it seems like a valid objection. If you take a very short click and EQ a couple 4-12dB notches below 200hz I think you could hear a clear difference. I don't think it's worth giving up room EQ but I don't see that it's an imaginary phenomenon.

I was thinking you could use a transient detector (been a thing with studio effects for a long time) to mix the EQ in and out, but the problem with that is if there are any steady tones (bass notes) going on behind the drums, you would hear extra thump along with the actual transient. I am not sure if there is a straightforward fix for this kind of thing, but it might be an improvement, another partial compromise.

 

[gnarly]

20ms is (as a rule of thumb) the shortest sound where it's perceptible as tonal, i.e. not just a click. Bass notes are pretty much always much longer than this... load up a sample in Audacity or something and trim it to 20ms, you'll see what I mean.

Well, i still don't think notes, however long or short they are, is the loss of impact Paul Hales used in his explanation.
Or what I've been referring to...which is a major transient bass attack...a hit to the chest .
The whack of a kick drum...that Paul says delivers 90% of its energy in 20ms.

I'm not saying don't knock down room-modes with EQ/DSP....I'm just saying like others, don't think it comes without price.
While I'm at it, I think we need to remember, multi-sub is a kissing-cousin in terms of mudding up bass transients.....

 

[janbth]

Thanks, super helpful and I think seeing this makes the argument a lot more intuitive, and I think establishes that the objection to EQ is at least plausible.
…
However, I do think the idea of losing impact from the room EQ is pretty plausible at this point, it seems like a valid objection.

It may seem plausible, but it’s not the case, at least not in my room. The boom coming from a nasty room mode is something you want to get rid of. So do we lose transient attack if we apply a PEQ filter to cut down the mode? In my experience with EQing down peaks in my own room, no, but if you are worried, listen to some burst test tones, like for example from this post:
https://www.audiosciencereview.com/...luation-of-room-modes-and-other-things.29310/
I find that the bursts (i.e. transient attack) sounds right - even at the cut down frequencies.

 

[janbth]

While I'm at it, I think we need to remember, multi-sub is a kissing-cousin in terms of mudding up bass transients.....

This is a strange thing to say. Multi-sub is a well known technique to handle room modes, that works very well. I have used a multi-sub technique myself with great success. I had deep, tight, slammy bass, so no, there is no mudding up. I’m currently using Dirac ART with multiple subs - even better!

 

[amirm]

So it is clear, benefits of Room EQ have been proven in controlled listening tests:

The lowest rated EQ systems lost because they tried to flatten the bass and or/introduce a dip.

 

[andyc56]

So it is clear, benefits of Room EQ have been proven in controlled listening tests:

This is the spinorama of the B&W 802N they used in the evaluation.

 

[Few]

I will have to watch the video…

Thanks for taking the time to listen to Hales’ argument. Greatly appreciated.

 

[Matt_Holland]

So it is clear, benefits of Room EQ have been proven in controlled listening tests:

The lowest rated EQ systems lost because they tried to flatten the bass and or/introduce a dip.

So the more EQ, the less the sound was preferred?
Why would an EQ system introduce a dip?

 

[amirm]

So the more EQ, the less the sound was preferred?

??? No. Those graphs were for different EQ systems. Two of them actually made things worse. Three others made things better. Key finding was that you had to smooth the response *and* dial in a bass boost to have that slope.

Why would an EQ system introduce a dip?

You can blame the person behind THX lettering for convincing that Audyssey EQ system that is shipped in millions of AVRs should have the so called "BBC" dip. The pro version lets you defeat that but the consumer one does not.

 

[Few]

If someone has a fairly focused perception question they'd like to study, such as what affects the perceived "impact" of a reproduced drum strike, I'm not confident that unfocused "What do you prefer?" studies yield the most helpful data. A listener whose tastes make bass impact a very low priority may not notice impact degradation. I'd be more interested in the result of a study that asks listeners to focus their attention on "impact," perhaps with a live drum listener-calibration step.

 

[Flaesh]

I'd be more interested in the result of a study that

This seems difficult to study correctly. Unless you compare the DBA with just the front array, equalized to the same frequency response.

Old Toole's pic:

But to what extent do we NOT hear direct sound at low frequencies?

 

[Matt34]

I got to listen to both parts of the interview and really enjoyed his take on what he perceives as an over-reliance on room correction by some to fix bad speaker design. and acoustically poor listening environments.

• Speaker problems → speaker design
• Room problems → room acoustics
• Low-frequency room modes → subwoofer placement + DSP
• Electronic problems → electronics
• Speaker/sub integration → DSP can be extremely useful

That's actually a much more defensible position than simply being "anti-room correction."

 

[Matt34]

If someone has a fairly focused perception question they'd like to study, such as what affects the perceived "impact" of a reproduced drum strike, I'm not confident that unfocused "What do you prefer?" studies yield the most helpful data. A listener whose tastes make bass impact a very low priority may not notice impact degradation. I'd be more interested in the result of a study that asks listeners to focus their attention on "impact," perhaps with a live drum listener-calibration step.

This has been a recent revelation for me (past couple of years) when I introduced Power Sound Audio subs with B&C pro audio drivers into my system.

 

[GL1zdA]

I'm a bit late to the party, but I've listened to the podcast yesterday. Here are his five-point problem with room correction:

1. Room EQ can't correct spatially variant signals. Can't have different processing for each chair in the room.
2. Room EQ can't correct both sustained signals and transients signals. You can't fix both of those separately.
3. Microphones do not detect the sound field the same way humans do. Microphones can't distinguish between reverberant, reflective energy and the direct energy.
4. Room EQ can't change the acoustic properties of the room or the dimensions of the room.
5. Room EQ can't EQ the reverberant field and direct sound independently.

I don't think anyone disagrees with point 1, that was the one of the Dr. Welti findings and the solution was multiple subwoofers. Or Dirac ART and Trinnov WaveForming take it further to active cancellation. I'm not familiar with what Room EQ systems, but it would be naive if they tried correct the average and this was his argument against Room EQ.

For point 2, I don't think anyone tries to fix them separately, EQ does the cut for both and the resulting lack of bass is addressed by the downward-sloped target curve.

While point 3 is true, I don't really see the problem here, since for useful measurements we capture the sound in some time window and can get this information back.

I don't really get point 4, it's empty for me. It's like saying bass traps can't fix a loudspeaker's response. Sure, they can't, but that doesn't mean they can't change what arrives at the listening position.

Point 5 can be generalized to "equalization cannot change loudspeaker directivity" and that's what both Paul Hales and Dr. Toole advocate: get better loudspeakers.

For me it really boil down to the question "will the transients be affected if we use Room EQ below the Schroeder's frequency targeting the Harman curve". Or maybe it's "can I skip Room EQ if I don't listen to music with sustained bass notes". I glanced over Toole/Olive paper "The Modification of Timbre by Resonances: Perception and Measurement" and maybe that's his answer?

 

[Flaesh]

"can I skip Room EQ if I don't listen to music with sustained bass notes"

IMO the lower organ pipe is less demanding in terms of low frequency reproduction quality than the staccato 40-50 Hz. ?))

I don't use an equalizer for room correction. Link below.