Measure your C50 in REW and tell us how your bass sounds!

[kawauso]

The timing of the spikes tells you the delay, and from the delay you can work out the distance. Then you look around your room and ask yourself which surface is responsible for the reflection. All described in the REW eBook.

ASR people must be sick of me pushing that book by now. But that's why the book was written - these questions come up really often.

And regarding the eBook:
I did learn some REW operations I didn’t know before, and I appreciate that.
But for features that I don’t use regularly, I feel I’m likely to forget the workflow again after a while.
There isn’t much I can actually apply in day-to-day use.

 

[kawauso]

I overlaid the C50 results for ART, Bass Control, and Room Correction in the bedroom.
The improvement with ART was expected, but BC actually went in the opposite direction.

EDT

T20

FR

 

[Keith_W]

Sure, I understand how to identify the surface geometrically.
That part is clear enough.

But what I'm still missing is the actionable part:

Even if I know
“this spike comes from the right wall 1.7m away,”
I still don’t know:

• Should I absorb it?
• Should I diffuse it?
• Should I ignore it?
• How strong does a reflection have to be before it's actually a problem?
• How much reduction is the target?
• What counts as “acceptable” in a small domestic room?

Your eBook explains how to inspect ETC beautifully,
but it does not really explain what to do with the information
or what thresholds matter in practice,
at least for typical living-room environments.

So I can identify reflections, yes.
But the step from “I know the delay” to “I know the correct treatment”
is not explained, and that’s the part I was hoping to understand.

Well it's a book about interpreting measurements. Not about what to do with what you see If I wrote a book like that, it would be freaking long and I would have to charge for it!

Re: a spike in the ETC. I think the book says that it's a problem if it's less than -15dB to the main impulse within the first 20ms. It will have the effect of smearing the main signal if it's loud and early enough. I just checked because you said I didn't mention thresholds - for a moment I thought I screwed up and forgot to include that information. Turns out I didn't! I remember creating that diagram! *smack*!!

Anyway, your options are:

- toe speakers in or choose speakers with narrower directivity. Sometimes Ishiwata toe-in can help.
- move speakers into a larger room or change orientation of speakers - for e.g. repositioning speakers against the long wall will help reduce and delay side wall reflections. I know this sounds crazy, but you could even experiment with an Ambiophonics type setup like STC's system - link. Scroll down a bit and see how his speakers are placed side by side.
- absorption vs. diffusion - lots of threads on ASR and it's somewhat controversial. IMO you should look at the RT60 (see the section on how to take and interpret a RT60 measurement). If it turns out that you would benefit from absorption, then install absorbers. If the RT60 is already pretty low, then diffusers.

Sometimes there's nothing you can do - can't move house, can't buy new speakers, have to obey wife, etc. Then you'll have to live with it. Physical problems require physical solutions, that's all there is to it.

 

[Keith_W]

The improvement with ART was expected, but BC actually went in the opposite direction.

The step response will tell you why. The eBook has a nice example of DLBC screwing up the step response by sending subwoofer delay in the wrong direction. I was kind of polite in the eBook and I said "I don't know if it's sampling bias" but I have seen enough DLBC step responses to know that it ALWAYS seems to do this. The only thing preventing me from calling DLBC a scam is the fact that I haven't personally tested it, so I don't know if 100% of the DLBC step responses I have examined are from people who are using it incorrectly, or whether the algorithm itself is at fault. I look at almost every MDAT that is posted on ASR, so I have examined maybe a dozen examples.

 

[kawauso]

The step response will tell you why. The eBook has a nice example of DLBC screwing up the step response by sending subwoofer delay in the wrong direction. I was kind of polite in the eBook and I said "I don't know if it's sampling bias" but I have seen enough DLBC step responses to know that it ALWAYS seems to do this. The only thing preventing me from calling DLBC a scam is the fact that I haven't personally tested it, so I don't know if 100% of the DLBC step responses I have examined are from people who are using it incorrectly, or whether the algorithm itself is at fault. I look at almost every MDAT that is posted on ASR, so I have examined maybe a dozen examples.

Thank you for your advice, but I’ll have to decline.
The choice of speakers, toe-in, diffusion, and absorption was already made carefully.
My goal is simply to enjoy good sound — not to chase lower reflections just because the ETC graph says so.

You mentioned “can’t disobey the wife,” but I am the wife — and the queen.

I also checked the step response. Indeed, BC shows a double peak,
but that might be an intentional offset to optimize other parameters.
It’s impossible to judge performance based on this alone.

You said you haven’t personally tested DLBC — have you actually listened to it?
It’s difficult to outperform even for experienced EQ users.
If you haven’t tried it yet, I recommend testing it.
The PC version requires no additional hardware and is free to experiment with.

 

[Flaesh]

Well it's a book about interpreting measurements. Not about what to do with what you see..

 

[Keith_W]

You mentioned “can’t disobey the wife,” but I am the wife — and the queen.

Wakarimashita, kawauso-san!

Anyway, those measurements are just a guide. I might have said somewhere in the book that if you don't hear a problem, then there is no problem. In the end it's what makes you happy, not what the measurements say.

You said you haven’t personally tested DLBC — have you actually listened to it?

Hai! I've heard DLBC, ART, Waveforming, MSO, and all sorts of DSP calibrations, including the DIY and semi-DIY options. It's hard to separate what is DSP and what isn't if there is no before-after blind comparison. And of course, half of the result depends on how well the user used the tool. Maybe all of the result if we are talking manual DSP.

View attachment 493723

There's something very odd and very ugly going on there. But it's kind of a bit too smooshed together to see clearly. And I don't know what the two measurements are, since they are not labelled. The yellow one does look similar to other DLBC step responses I have examined previously, and in line with what we see in your clarity measurement. I don't know what the blue line is, I don't think it's ART since ART usually places the subwoofer impulse before the main impulse.

 

[kawauso]

Wakarimashita, kawauso-san!

Anyway, those measurements are just a guide. I might have said somewhere in the book that if you don't hear a problem, then there is no problem. In the end it's what makes you happy, not what the measurements say.



Hai! I've heard DLBC, ART, Waveforming, MSO, and all sorts of DSP calibrations, including the DIY and semi-DIY options. It's hard to separate what is DSP and what isn't if there is no before-after blind comparison. And of course, half of the result depends on how well the user used the tool. Maybe all of the result if we are talking manual DSP.



There's something very odd and very ugly going on there. But it's kind of a bit too smooshed together to see clearly. And I don't know what the two measurements are, since they are not labelled. The yellow one does look similar to other DLBC step responses I have examined previously, and in line with what we see in your clarity measurement. I don't know what the blue line is, I don't think it's ART since ART usually places the subwoofer impulse before the main impulse.

I forgot to add the labels earlier — the blue trace is RC and the yellow trace is BC.

I don’t intend to ignore measurements, but I also don’t want to be controlled by them.
I’m trying to decide for myself which metrics matter and which ones I shouldn’t worry too much about.

For example, with speakers, I used to focus heavily on Spinorama scores.
Recently, though, I’ve realized that I prefer consistent directivity over scores that rely heavily on EIR weighting.
While ±60–70 degrees of wide dispersion is generally popular, I find that a narrower, sharper pattern — around ±50 degrees or even less — is more comfortable for my taste.

As for frequency response, I used to correct all the way up into the high frequencies because small bumps bothered me,
but now I feel it’s more natural to leave anything above around 300 Hz to the room’s reflections.

Regarding Dirac, RC and BC don’t look that different from other auto-EQ systems like ARC or Audyssey in terms of FR,
yet they clearly sound better to me.
I suspect this is because I prefer when the phase is consistent across channels.

 

[Flaesh]

This thread isn't about frequency response.
May C(D) metrics using RC\BC be compared, all other things being equal?))

 

[kawauso]

We went slightly off-topic.
Here are the measurements from the same bedroom in a 2.2-channel setup.
It’s a small Japanese-style room with very little mid–high frequency reverberation.
These are the graphs for RC (Dirac Basic), BC (multi-sub + crossover), and ART.

Subjectively, RC sounded a bit clearer in the bass than BC,
but I listened after looking at these graphs, so it was probably bias.
I’m not confident I could distinguish them in a blind test.
ART, on the other hand, is clearly more defined.

RC

BC

ART

C50

 

[audiofooled]

A test you can do on headphones is to play 16th second bursts with 16th second rests between them at various frequencies. See how low you can go before you really can't make out the starts and stops of the bass tone clearly. If by bass we are exlcuding sub-bass and only talking about frequencies above 60Hz, you'll find that you can clearly hear 16th notes all through the bass range on headphones. This is not the case in most people's listening rooms. The rests between the 16th notes gets filled with reflected sound that makes the notes blur together to various amounts perceptually. This is a considerable problem in many people's listening rooms all the way up to over 600 Hz. Above that most people's rooms will clearly reproduce 16th note bursts and rests, so we're not so concerned with that. Up in that range it's more issues with tonality and imaging.
A 16th note is 62.5 millliseconds. C50 is dealing with 50 milliseconds so that's pretty close, and the perceptual testing I've done on myself with test signals suggests that if the energy in that period after the burst corresponds to about a 10 dB score on the C50 in the bass range, the bass will sound detailed and "quick." A lot of people's scores on this thread are way beyond that. But there are many people who have frequency ranges that are falling well below that, and when I have them play a test sweep of bass frequency bursts I can see and hear that they are having many problem areas at various frequencies.

You can listen to these bursts here: https://www.audiocheck.net/audiotests_matt.php

I just listened to it on my system and I've got a few frequency bands where the bursts get mangled. I can still hear the high frequency ticks from the start of the impulses, but the fundamental tone is blurred. This is a better, more detailed test than the C50 in REW because it lets you hear where you're having problems. Unfortunately it doesn't tell you what frequency you are at in any part of the test.
You can generate these bursts in REW at any frequency you want and find out if there are any problem frequency bands in your room.
I've been working on software that will analyze the burst sweep and chart out a detailed volume level chart that is color coded based on the difference in volume between the peak burst level and the average level in the rest between bursts. You can select any part of the chart and listen to the bursts in that zone, comparing a good measuring area to a bad measuring area. What I can tell you is that if you have a room where the C50 measurement is over 10 through the entire bass range there won't be many audible problems in terms of bass clarity.
Some people may argue that music doesn't use a lot of 16 note speed below 100 Hz. I think that's more or less true. But we can certainly hear it, so I think it makes sense that it should be clearly reproduced at our listening position.

Why not burst decay then? I believe that in rooms, when it comes to low frequencies, the lower the frequency, the more energy, the more control you need.

Headphones, point blank:

L+R+sub, MLP, 3,5m distance:

RT60 (untreated room):

Vector average:

RT60:

The fourth problem: you would think that higher clarity would be better, right? Not so. If you load up your room with acoustic treatment, you will reduce the "late" part and get a very high "clarity" score. Yet we know that overtreating rooms and pushing the RT60 down below 200ms actually reduces speech clarity (I can supply a Toole reference if needed). So, low clarity is bad, high clarity is bad ... what is the ideal "clarity" range? Nobody really knows.

+1

In general, if having a normal conversation in the room sounds natural, we also use brain processing for speech intelligibility in sort of hearing through the room when it comes to reproduced dialogue. If the loudspeakers are well behaved to the point that nothing is really sticking out as objectionable, I believe it can also sound natural.

On the other hand, low bass are really frequencies that cannot fit. In my room, 16 Hz suckout is not even related to this room, but larger rooms underneath (garage, basement and workshop) that resonate at that frequency directly underneath the system. Don't ask how I know.

 

[Tim Link]

Why not burst decay then? I believe that in rooms, when it comes to low frequencies, the lower the frequency, the more energy, the more control you need.

Headphones, point blank:

View attachment 493910

L+R+sub, MLP, 3,5m distance:
View attachment 493911

RT60 (untreated room):

View attachment 493912

Vector average:

View attachment 493913

RT60:

View attachment 493914



+1

In general, if having a normal conversation in the room sounds natural, we also use brain processing for speech intelligibility in sort of hearing through the room when it comes to reproduced dialogue. If the loudspeakers are well behaved to the point that nothing is really sticking out as objectionable, I believe it can also sound natural.

On the other hand, low bass are really frequencies that cannot fit. In my room, 16 Hz suckout is not even related to this room, but larger rooms underneath (garage, basement and workshop) that resonate at that frequency directly underneath the system. Don't ask how I know.

Burst decay is great. I find various spectrograms very informative. The trick for me is getting the settings right so that what I see correlates to what I hear.

 

[Tim Link]

The fourth problem: you would think that higher clarity would be better, right? Not so. If you load up your room with acoustic treatment, you will reduce the "late" part and get a very high "clarity" score. Yet we know that overtreating rooms and pushing the RT60 down below 200ms actually reduces speech clarity (I can supply a Toole reference if needed). So, low clarity is bad, high clarity is bad ... what is the ideal "clarity" range? Nobody really knows.

There are some reasonably good ideas about room clarity in the lower midrange and bass. Turns out this correlates pretty well to about a 300 ms RT60, which is generally agreed to be a pretty good target for home sized audio playback spaces. Maybe a little faster for home theater surround sound setups. Maybe a little longer for people who prefer a more lively sound. I think it also correlates well with a C50 measurement of at least 10, because it can be demonstrated reliably that C50 less than that can audibly smear transients in the lower midrange and upper bass, creating a less satifisfying stereo listening experience even though speech clarity may remain high enough to easily understand.
The Master Handbook of Acoustics goes over much of this. Coherent reflections within the time fusion window , about the first 50 ms, can actually aid in our ability to hear speech, if for no other reason than they make it sound louder. These reflectins can color the tone though, so it's best if no single one of these reflections is too loud compared to the direct signal, and also there should be plenty of them, averaging about 1 per millisecond, but not evenly spaced apart as that will cause it's own weird audible issues. We can tolerate and perceptually adjust for tonal shifts to quite an extent, but it's nicer if they aren't so pronounced. For stereo imaging it's often suggested that it's best if there is a delay of 10 ms before the first reflection reaches us, and it's also best to minimize crosstalk reflections. So we'd like a lot of weak, coherent reflections that come from the same side of the room as the speaker that generated them, and come from different lateral directions, which will also help with a sense of envelopment and further ease our process of tonal adaptation to the space.
Breaking up large flat surfaces in an optimized way using a combination of absorption and diffusion can go a long way.
This is an interesting presentation if you haven't watched it already

 

[audiofooled]

Burst decay is great. I find various spectrograms very informative. The trick for me is getting the settings right so that what I see correlates to what I hear.

IMO/IME no single measurement would tell you what you hear, even though there could be many that suggest so. Apart from many strategies for not supporting the room resonances, time domain performance could be well correlated. For my system, the more important ones I show here:

Stereo Bass using subwoofers

Um, sound pressure is proportional to f^2 * volume velocity. There's 2 accelerations involved, cone displacement and microphone displacement, reacting to pressure, so from end to end, it's f^2 * displacement. That's why speakers can be flat, after all, the second-order lowpass above resonance...

www.audiosciencereview.com

also here:

Frontal reflections, depth of soundstage, and dipoles

As @tmuikku has suggested in various posts/threads I spent this afternoon listening to musical selections and test tracks at different positions around the room. Room is 19'2"W x 24'D x 8.5'H and purpose built/professionally designed. Front baffle of the speakers are 3'6" from the front wall and...

www.audiosciencereview.com

A Broad Discussion of Speakers with Major Audio Luminaries

The field of audio is small. “String of temporary opinions” sadly largely covers it, like palaeoanthropology from where it comes. There are exceptions, e.g. directivity of loudspeakers, the early Nordic standards on FR in reproduction, momentary hearing thresholds etc. However, we rely so much...

www.audiosciencereview.com

Here, albeit no mic involved, I try to demonstrate how the shape of wavefront and floor bounce control can be an important factor, basically it provides a visual of the attack and decay:

Bass direction is audible

I would love to hear ASR's opinion about this. https://www.audiosciencereview.com/forum/index.php?threads/visualizing-how-different-loudspeaker-lf-directivity-patterns-couple-to-room-modes.45518/ I don't know why NTK didn't provide this link))

www.audiosciencereview.com

Needles to say, perception is complicated, taking into account that brain works on perceived loudness, not SPL. Pressure is only a part of the equation.