or Connect
AVS › AVS Forum › Audio › Audio theory, Setup and Chat › Simplified REW Setup and Use (USB Mic & HDMI Connection) Including Measurement Techniques and How To Interpret Graphs
New Posts  All Forums:Forum Nav:

Simplified REW Setup and Use (USB Mic & HDMI Connection) Including Measurement Techniques and How... - Page 231

post #6901 of 9520
Quote:
Originally Posted by kbarnes701 View Post

I read that professional sound engineers are happy with +/- 5dB and concluded that, if so, it's probably more than good enough for me too. I don't want to chase perfect graphs - I am just not that interested (heresy I know). I just want very good SQ in my room.

And you're on THIS thread? Says the man with the tuned membrane trap and the gubbins....smile.gif

(emoticon may be in limbo as the thread continues)

Gear mentioned in this thread:

post #6902 of 9520
Quote:
Originally Posted by AustinJerry View Post
 
Quote:
Originally Posted by jim19611961 View Post




Here is a very old FR of the same response in 1/3 and 1/24th.

The null at 33hz is completely smoothed over at 1/3rd. And the peak at 64hz, the same. 200hz, the same. I could hear EASILY these phenomena.

But my greater point is at 1/3rd (1/6th not being much better), you wouldn't be aware of these at all.

 

I think we understand, Jim.  I would hope that we all look at the 1/24 smoothed plot to identify areas that need addressing.  However, in your example above, other than the issue at~35Hz, I don't see any other glaring problems.  Of course, that would depend on one's objectives, I guess.

 

And we always use no smoothing for the <300Hz anyway... not sure what is going on here.

post #6903 of 9520
Quote:
Originally Posted by jim19611961 View Post
 
Quote:
Originally Posted by kbarnes701 View Post

I’d be happy with either!  I was under the impression that 1/6th more accurately reflects what we actually hear. Is this not so?

In terms of the general tone of the room, yes. But peak and valleys can a very narrow Q, and therefore not show up on overly smoothed graphs.

If you had a narrow peak of +6db @ 440hz (A4), when that note comes along, it will suddenly seem much louder than it should. There is no way you wont. But a overly smoothed plot may tell you its only a 1 or 2db peak or miss it entirely.

I think we are misunderstanding each other. I use no smoothing at all <300Hz and 1/24th >300Hz for analysis. I only use 1/6th after the analysis etc has been done, to get a 'feel' for what I have achieved, in a way that my ears will hear it. Is this not correct?

post #6904 of 9520
Quote:
Originally Posted by jim19611961 View Post
 
Quote:
Originally Posted by kbarnes701 View Post

I read that professional sound engineers are happy with +/- 5dB and concluded that, if so, it's probably more than good enough for me too. I don't want to chase perfect graphs - I am just not that interested (heresy I know). I just want very good SQ in my room.

Over at GS, there are a lot of sound engineers. And although I cant say ive read what you did, what I have read is that they agree on is +/-5db at 1/24th oct smoothing. +/-3db at 1/3rd oct smoothing is another you will see. But the smoothing qualifier is a must.

But even these generalizations are misnomers because where the +'s and -'s are must be considered. Obviously, if you were at +5db at 1k, and -5db at 100hz and 10k, that would sound horrible.

 

Yes, good point. I am +/-3dB at 1/3rd smoothing. Is that good enough then?  I am starting to get confused. And should that be for individual channels, combined L+R+subs or what?

post #6905 of 9520
Quote:
Originally Posted by sdrucker View Post
 
Quote:
Originally Posted by kbarnes701 View Post

I read that professional sound engineers are happy with +/- 5dB and concluded that, if so, it's probably more than good enough for me too. I don't want to chase perfect graphs - I am just not that interested (heresy I know). I just want very good SQ in my room.

And you're on THIS thread? Says the man with the tuned membrane trap and the gubbins....smile.gif

(emoticon may be in limbo as the thread continues)

 

Hehehe. I am not as concerned about the ultimate graph as some are. I am full of admiration for those who are, but it isn't for me. Once we get below the threshold of audibility my interest disappears entirely I am afraid. I decided against tuned membrane traps, as anyone would, after seeing Jerry's results. I did like the look of limp membrane traps though - and if I was starting again I think they are what I might choose to use. 

post #6906 of 9520
Quote:
Originally Posted by kbarnes701 View Post

Yes but surely those peaks and valleys are entirely inaudible?

http://www.audioholics.com/room-acoustics/human-hearing-amplitude-sensitivity-part-1



The finding by these various studies suggest we can discriminate between 0.25 and 3db differences. And all but one suggest in the 1db - 1.5db range.
post #6907 of 9520
Quote:
Originally Posted by kbarnes701 View Post

Yes, good point. I am +/-3dB at 1/3rd smoothing. Is that good enough then?  I am starting to get confused. And should that be for individual channels, combined L+R+subs or what?

Again, I am not hear to tell anyone whats good enough for them.

These guidelines should be applied to individual channel responses (L, R, L+sub, R+sub). Combining L+R naturally intruduces a smoothing or averaging effect upon them. But many treat the bass region differently and apply different criteria to it. For one, many low bass instruments are in mono anyway, so combined channel responses sometimes emulate whats going on anyway pretty well. Lower bass frequencies lose directionality as well.

I am not saying L+R cant be useful. But you may have to apply different goal constraints upon it.

But personally, I find +/-5db at 1/24th oct smoothing more useful and telling than +/-3db at 1/3rd oct smoothing. But many here, including self, elevate the bass region, so these +/- factors have to be modified.
Edited by jim19611961 - 11/14/13 at 3:12pm
post #6908 of 9520
Quote:
Originally Posted by jim19611961 View Post



Here is a very old FR of the same response in 1/3 and 1/24th.

The null at 33hz is completely smoothed over at 1/3rd. And the peak at 64hz, the same. 200hz, the same. I could hear EASILY these phenomena.

But my greater point is at 1/3rd (1/6th not being much better), you wouldn't be aware of these at all.

Your points are well founded in the low bass area. Simplistically speaking if one divides up their analysis into >250Hz and below <250Hz (corresponding to modal / specular regions) then what I stated was true above 250Hz. I will edit my post to that effect.

Do you think we can hear peaks and dips which show up at 1/24th resolution at say 800Hz? I don't think so.

As frequency goes down our frequency resolution increases. I'm not convinced it's as much as 1/24th octave though. Certainly 1/12th octave.

What's more we really don't hear the nulls like we hear the peaks. What we hear is more like a the arc drawn by say a swing bridge or cable car between two mountains.
post #6909 of 9520
Quote:
Originally Posted by AustinJerry View Post

Nyal, can you clarify what you mean by "not valid"?  In the example below, are you saying that combined L+R measurement has no useful information?



In the low bass, say <250Hz or where ever there might be the same signal playing from more than one speaker then L+R or L+R+C+Sub is valid.

Differences in mic to speaker distances will cause comb filtering when playing the same signal through multiple speakers, especially as the wavelengths get very short.
post #6910 of 9520
Quote:
Originally Posted by Nyal Mellor View Post


Do you think we can hear peaks and dips which show up at 1/24th resolution at say 800Hz? I don't think so.


Reisz, Dimmick & Olson, Atal, Jestaedt, Toole and Olive would seem to suggest we do, yes.
post #6911 of 9520
Quote:
Originally Posted by jim19611961 View Post

http://www.audioholics.com/room-acoustics/human-hearing-amplitude-sensitivity-part-1



The finding by these various studies suggest we can discriminate between 0.25 and 3db differences. And all but one suggest in the 1db - 1.5db range.

Most of those studies were done with headphones or in anechoic chambers. In rooms with a multitude of reflections I am convinced we hear things very different. I don't think we really hear the peak and dip pattens (comb filtering) that in a measurement is caused by superposition of direct sound and out of phase reflections. I think what we hear is more a set of snapshots of the sound (both direct and reflected), and the phase information that causes the peaks and dips on the measurements doesn't really come into things.

This is well summarized by this great quote from Benade, as requoted by Toole:

“The auditory system combines the information contained a set of reduplicated sound sequences (authors note – i.e. the direct sound and its reflections) and hears them as if they were a single entity, provided:

a. that these sequences are reasonably similar in their spectral and temporal patterns,

and

b. that most of them arrive within a time interval of 40 ms following the arrival of the first member of the set.

The singly perceived composite entity represents the accumulated information about the acoustical features (tone color, articulation, etc) shared by the set of signals. It is heard as though all the later arrivals were piled upon the first one without any delay – that is, the perceived time of arrival of the entire set is the physical instant at which the earliest member arrived (authors note – this is known as the precedence effect).

The loudness of the perceived sound is augmented above that of the first arrival by the accumulated contributions from the later arrivals.

The apparent position of the source of the composite sound coincides with the position of the source of the first-arriving member of the set, regardless of the physical direction from which the later arrivals may be coming.”
post #6912 of 9520
Quote:
Originally Posted by Nyal Mellor View Post

Most of those studies were done with headphones or in anechoic chambers. In rooms with a multitude of reflections I am convinced we hear things very different. I don't think we really hear the peak and dip pattens (comb filtering) that in a measurement is caused by superposition of direct sound and out of phase reflections. I think what we hear is more a set of snapshots of the sound (both direct and reflected), and the phase information that causes the peaks and dips on the measurements doesn't really come into things.

This is well summarized by this great quote from Benade, as requoted by Toole:

“The auditory system combines the information contained a set of reduplicated sound sequences (authors note – i.e. the direct sound and its reflections) and hears them as if they were a single entity, provided:

a. that these sequences are reasonably similar in their spectral and temporal patterns,

and

b. that most of them arrive within a time interval of 40 ms following the arrival of the first member of the set.

The singly perceived composite entity represents the accumulated information about the acoustical features (tone color, articulation, etc) shared by the set of signals. It is heard as though all the later arrivals were piled upon the first one without any delay – that is, the perceived time of arrival of the entire set is the physical instant at which the earliest member arrived (authors note – this is known as the precedence effect).

The loudness of the perceived sound is augmented above that of the first arrival by the accumulated contributions from the later arrivals.

The apparent position of the source of the composite sound coincides with the position of the source of the first-arriving member of the set, regardless of the physical direction from which the later arrivals may be coming.”

I agree that a reflective room makes a difference in discrimination ability. But one of the goals in many/most acoustically treated rooms is to diminish these very reflections you speak of. I would argue that reflections of the order of -20db or -30db are effectively taken out of the equation, and thus couldn't be considered relevant to the point your making.
post #6913 of 9520
More here: http://www.moultonlabs.com/more/about_comb_filtering_phase_shift_and_polarity_reversal/P1/

This is just my position on things. I think sometimes we chase things in the measurements that aren't really of consequence to what we hear.
post #6914 of 9520
Quote:
Originally Posted by Nyal Mellor View Post

In the low bass, say <250Hz or where ever there might be the same signal playing from more than one speaker then L+R or L+R+C+Sub is valid.

Warning: heresy may be coming. This is just a thought experiment, which is either stupid or innovative:
The latter is very interesting to me, since I have powered L/C/R, with Mythos ST main and a CS-8080 HD center channel. Even with 12 db/octave filtering below the crossover, there may still be some useful output for them at, say, 70 or 75 Hz with an 80 Hz crossover, as crossovers aren't brick walls. These speakers have built-in powered "woofers", and while I don't pretend that they're subwoofers, if I'm trying to capture the whole picture of what's happening with my bass, I've wondered whether running an L/C/R+subs plot below, say, 150 to 250 Hz is helpful. Obviously not for two-channel listening, but if you listen to multichannel music, where a sense of bass slam in the center of the image is something I like, maybe it matters for accessing FR or spectrograms for the exclusive purpose of considering overall low/mid bass response.

I used to run L/C/R+Sub plots with OmniMic using one of the DSP modes back when, but as my knowledge grew and we collectively moved to REW, I backed away and used L/R+sub or individual speakers (as well as subs only) on FR plots in the no smoothing, 15-300 Hz range.

However, how you'd actually do this with REW/HDMI and the Multi CH signal input is a different issue.
post #6915 of 9520
Quote:
Originally Posted by Nyal Mellor View Post

More here: http://www.moultonlabs.com/more/about_comb_filtering_phase_shift_and_polarity_reversal/P1/

This is just my position on things. I think sometimes we chase things in the measurements that aren't really of consequence to what we hear.

I think its through sharing our positions that we come to know with greater clarity this very distinction. smile.gif
post #6916 of 9520
Quote:
Originally Posted by djbluemax1 View Post

So REW's default settings of a 40db range and 'Match Scale To Peak' seem to provide decent graphs, but here's the kicker. As we've noticed, viewing the same graph with a different top and bottom range shows different decay, even with the same 40db range.

Doesn't that mean that graphs with significant peaks (raising the top end and thus, range of the scale) will produce different results than a more even graph without huge peaks?

Perhaps we can just request that folks always post a Waterfall with a Spectrogram? Although the Spectrogram 'appears' a little easier to read, it's the consistency of the results that I'm wondering about. At least with the waterfalls, as long as there's at least a 40db range from the top of the measurement (the mountain, not the window) to the bottom, we can see if it's running into the noise floor down low, and if the measurements should be redone at higher SPLs. The results though, are more consistently read.

I think the general idea is to see the decay from the peak amplitude. This is going to be subject to variance depending on the smoothness of the measured room and also on how the level has been set for the bass frequencies. Consider a smooth FR with a 10dB peak at 100hz, if I was set the level so that the 10dB peak matched the rest of the FR then the spectrogram would be accurate imo because I am measuring from that 10dB peak. On the other hand, if I was to set the levels so that the rest of the bass region matched the entire FR allowing that 10dB peak to be a peak over the FR, then the spectrogram would be inaccurate in this instance imo, because you would be measuring from a peak that is 10dB above everything else. It would 'hide' the issues at all frequencies except the 10dB peak.

Does that make sense for anyone other then myself?

With the waterfalls graphs suggested to be set @ 45dB for the low limit, this allows a little extra detail to be shown if people measure at a sufficient loudness.

Quote:
Originally Posted by jim19611961 View Post

One reason I brought up the issue was I wanted to see what others are getting to improve my understanding.

Here is mine Jim.



For completeness, here it is out to 70ms.



Quote:
Originally Posted by jim19611961 View Post

The black curve (red = now) is where I once was and my null at 225hz was primarily a floor bounce issue. Honestly, this issue was one of the most troublesome and time consuming problems to fix of ALL my room issues. In my case, optimizing speaker placement alone wasn't enough to deal with it. IMO, thick carpet and pads dont do much in the 150-250hz range.

Here's a FR plot of mine.


I have an issue @ 250hz. Changing the placement only shifts the null up/down 50hz or so. It's been driving me crazy. Only you guys here could understand how excited I am to lay pink fluffy stuff on my floor and blast sine waves through the speakers biggrin.gif
Note: I have no dampening on the ceiling and only carpet and a rug on the floor.

Also, green and red are the left and right speakers with blue being the combined measurement. You can see the high frequency issues that arise from measuring multiple speakers.
post #6917 of 9520
Quote:
Originally Posted by Audionut11 View Post


You can see the high frequency issues that arise from measuring multiple speakers.

Thanks for sharing.

BTW, you can get the L+R out to 20k correct when the mic is EXACTLY the same distance from each source.

post #6918 of 9520
I had it about exact as I could make it. I have 3 kids under 6 (2 boys), so the listening position gets shifted whenever someone decides to use my chair as the command station!
post #6919 of 9520
Quote:
Originally Posted by jim19611961 View Post

Thanks for sharing.

BTW, you can get the L+R out to 20k correct when the mic is EXACTLY the same distance from each source.


Physical or in the AVR delay settings, or both?
post #6920 of 9520
Quote:
Originally Posted by Audionut11 View Post

I had it about exact as I could make it. I have 3 kids under 6 (2 boys), so the listening position gets shifted whenever someone decides to use my chair as the command station!

Unfortunately, even off by 1/4" will foul it up.
post #6921 of 9520
Quote:
Originally Posted by sdrucker View Post

Physical or in the AVR delay settings, or both?

Cant comment on the AVR aspect (dont have one), so I mean physically. Omnimic makes this easy given its real time output. You just push the mic a hair at a time until the 15-20k part of the plot maxes out.
post #6922 of 9520
REW has a RTA also.
post #6923 of 9520
Quote:
Originally Posted by Audionut11 View Post

REW has a RTA also.

Haven't tried it, but that could work smile.gif
post #6924 of 9520
Quote:
Originally Posted by jim19611961 View Post
 
Quote:
Originally Posted by sdrucker View Post

Physical or in the AVR delay settings, or both?

Cant comment on the AVR aspect (dont have one), so I mean physically. Omnimic makes this easy given its real time output. You just push the mic a hair at a time until the 15-20k part of the plot maxes out.

 

Are you saying that if we see high frequency rolloff when measuring L+R that this could be the reason?  I hadn't heard that before...but then again, all of this about issues measuring L+R is fairly new to the thread AFAIK.

 

[Insert quote from Markus here where he mentioned it months ago and I just missed it!] :D

post #6925 of 9520
Quote:
Originally Posted by jkasanic View Post

Are you saying that if we see high frequency rolloff when measuring L+R that this could be the reason?  I hadn't heard that before...but then again, all of this about issues measuring L+R is fairly new to the thread AFAIK.

[Insert quote from Markus here where he mentioned it months ago and I just missed it!] biggrin.gif

Yes, that is exactly what I am saying. Do you see HF rolloff in my graph? smile.gif

http://www.avsforum.com/t/1449924/simplified-rew-setup-and-use-usb-mic-hdmi-connection-including-measurement-techniques-and-how-to-interpret-graphs/6900_50#post_23949465
post #6926 of 9520
Quote:
Originally Posted by kbarnes701 View Post

But for the 'end result' of what the room looks like, isn’t the R+L+Subs the right thing to do?
Your L/C/R speakers are placed at the 3 nulls of your 3rd width mode so that their interaction cancels/minimizes the first 5 width modes of your room: 54Hz, 108Hz, 161Hz, 215Hz and 269Hz. The 1st width mode is below the crossover point and being addressed by your dual subs.

How will you verify that the interaction between your L/C/R speakers is reducing the next 4 modes if you don't measure the interaction between your L/C/R speakers? Measuring L+R won't let you see that. Hence Nyal's suggestion to measure all 3 front speakers (and subs) to see what is happening in the modal range.
post #6927 of 9520
Keith, was it established the Herb is no longer shipping worldwide? If so I thought you might want to know to change this in post #10. Just happened to see it while reading for more info.
Quote:
This Mic is only $90.00 and he charges $9.00 for shipping. He also ships Worldwide.
post #6928 of 9520
Ok, so guide says match scale to peak selected. But also says scale top DB 99 and bottom db 59. Sorry for the confusion but does this mean to match scale to peak but shoot for 99db? Ok I tinkered with 99db and match scale to peak. So what if I had a 105db sweep? Would I just click 6db lower and run again? I am just trying to visualize something like a 106 or 86 db run and the selecting 99db?


Wow! the new settings are very cool, the "draw contours really makes it easier to see. Please I mean no challenging, Just curious. How come no 3D enhancement? It seems even easier to read with that engaged. Im sure there are legitimate reasons.
Edited by jlpowell84 - 11/14/13 at 6:19pm
post #6929 of 9520
With 'match top of scale to peak', you can ignore 'scale top' and 'scale bottom' unless you want to manually change things. If you want to see further down into the noise floor you should use 'scale range'. This won't change the top end but will simply extend the scale further into the noise floor.
post #6930 of 9520
Quote:
Originally Posted by jlpowell84 View Post

Ok, so guide says match scale to peak selected. But also says scale top DB 99 and bottom db 59. Sorry for the confusion but does this mean to match scale to peak but shoot for 99db? Ok I tinkered with 99db and match scale to peak. So what if I had a 105db sweep? Would I just click 6db lower and run again? I am just trying to visualize something like a 106 or 86 db run and the selecting 99db?


Wow! the new settings are very cool, the "draw contours really makes it easier to see. Please I mean no challenging, Just curious. How come no 3D enhancement? It seems even easier to read with that engaged. Im sure there are legitimate reasons.

You are over-thinking it. The 99 and 59 values are defaults built into the software. When you select an actual measurement, the upper limit is automatically set to the peak level in your measurement. The lower value is set 40dB lower. Does this make sense?
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Audio theory, Setup and Chat

Gear mentioned in this thread:

AVS › AVS Forum › Audio › Audio theory, Setup and Chat › Simplified REW Setup and Use (USB Mic & HDMI Connection) Including Measurement Techniques and How To Interpret Graphs