Control Room - target decay rates

[soundslikejoe]

I'm making plans for the treatment and final touches. My room is very empty and very live sounding.

When testing the room what are target ranges of decay rate?

I plan to mostly mix but will track some things. I don't think I work at high volume, probably between to 80-85db. Because of the tracking I want to retain some decay. I don't really like anechoic places.

[Ethan Winer]

RT60 is not totally valid in control rooms. The smaller the room, the less likely you'll ever have true reverb. But you can still measure it and adjust treatment accordingly. The size of the room dictates the "ideal" decay times. A room the size of a bedroom should aim for between 100 and 300 milliseconds. Just as important, the decay times should be as even as possible over the entire range of frequencies. As it happens, I'm about to finish a video showing the RT60 as measured in a room 16 by 11.5 by 8 feet. The attached graph shows the decay times in third-octave bands for the room empty, then with a bunch of bass traps and reflection absorbers, and then again with diffusors covering all the remaining surfaces. You can see the huge improvement in uniformity with the diffusors in place.

--Ethan

[soundslikejoe]

Cool post E.

I read some briefly about RT60 here; TSC Tech Talk | RT60 Calculator

I tried to use the calculator but it didn't work. (browser maybe)

My room is 12.9x14x8 (roughly)..... any idea about the RT60 value or links to more info/calculators.

[xaMdaM]

Just for accuracy... this RT60 calculator DOES work: Reverberation Time Calculator

[Glenn Kuras]

Originally Posted by Ethan Winer

RT60 is not totally valid in control rooms. The smaller the room, the less likely you'll ever have true reverb. But you can still measure it and adjust treatment accordingly. The size of the room dictates the "ideal" decay times. A room the size of a bedroom should aim for between 100 and 300 milliseconds. Just as important, the decay times should be as even as possible over the entire range of frequencies. As it happens, I'm about to finish a video showing the RT60 as measured in a room 16 by 11.5 by 8 feet. The attached graph shows the decay times in third-octave bands for the room empty, then with a bunch of bass traps and reflection absorbers, and then again with diffusors covering all the remaining surfaces. You can see the huge improvement in uniformity with the diffusors in place.

--Ethan

Really looking forward to the video Ethan.

[jwl]

I've seen a sneak preview of the video, it's great. I wonder if people realize how extraordinary the RT60 curve in the above graph is in general, but especially for a small room.

The key is to make the reverb time even at all frequencies, as has been said.

[Ethan Winer]

Originally Posted by jwl

I wonder if people realize how extraordinary the RT60 curve in the above graph is in general, but especially for a small room.

Indeed. I'll see if I can find a way to make an even bigger point about that.

--Ethan

[DanDan]

Well done Ethan. Your tests as always are a wake up call.
I look forward to seeing the video.
The results are indeed extraordinary. I don't have third octave, as I haven't moved on to FuzzMeasure 3. However, I note that I have never gotten reliable, repeatable EDT numbers in small rooms. We know that RT is not really defined in such small spaces.
I recommend that you check the room with other software. Maybe some friend locally has Dirac?
Perhaps REW excels at small room decay measurement?
Does REW create these RT graphs using multiple measurement locations?
Did you drive both speakers? I get better Waterfalls and Decay readings with double drive. Genelec have also done research indicating more even room stimulation using multiple Subs.
I am intrigued by the decrease in HF RT caused by the Diffusors. Could those HF figures have been flutter or other rings or pings?
How many Traps, and of what type? Ditto Diffusors.

The BBC used a criteria that no third octave RT should be more than 10% different from it's neighbours.
It is commonly accepted that the RT should increase towards LF.
I would be very wary of an even response in either the time or frequency domains.
I am thinking the room described would sound very bright, a great vocal room, but I would be a very wary of mixing in it. Can you show the Frequency response and Waterfall plse?
Best, DD

[Ethan Winer]

Originally Posted by DanDan

I note that I have never gotten reliable, repeatable EDT numbers in small rooms. We know that RT is not really defined in such small spaces.

Yes and No. I'm often the one to point out that small rooms don't have "real reverb" but rather a series of individually decaying reflections. However, even a small room has real reverb when totally empty.

A few weeks ago we did a similar test, which I used as the basis of this article:

Hearing is Believing

That article has no data though I did make measurements. Then I decided to make it into a video and retested again more carefully. Anyway, I just loaded up the original files and they show nearly identical changes in RT60. And it's the change that matters almost more than anything. I've never gotten reliable (read: repeatable or sensible) RT60 data from the ETF software, but REW is not only reasonable but repeatable. The main difference between my last test and the one posted above is the first test had no bass traps in the ceiling corners. But the RT60 differences are visually about the same as for this second test. The attached screen-cap is from our first test. It's not labeled, but the green is bass traps and bare walls, and the orange is after adding diffusors over the bare walls.

Does REW create these RT graphs using multiple measurement locations?

Well, you can measure in multiple places and view the results.

Did you drive both speakers? I get better Waterfalls and Decay readings with double drive.

Yes, I do most testing with both speakers fed the same source in mono.

I am intrigued by the decrease in HF RT caused by the Diffusors. Could those HF figures have been flutter or other rings or pings?
How many Traps, and of what type? Ditto Diffusors.

Rather than spend time to describe it all here, you'll see the setup in the video soon enough. But it makes sense that diffusors would not only even out the times but also lower them due to losses within the wells.

I am thinking the room described would sound very bright, a great vocal room, but I would be a very wary of mixing in it. Can you show the Frequency response and Waterfall plse?

The room as fully treated is incredibly clear and neutral, and fantastic for mixing. Again, you'll see not only all the data and photos of the setup, but be able to hear music recorded through microphones. And not only in the front of the room where RFZ reflections are absorbed, but in the rear too where you can compare diffusors versus bare walls.

--Ethan

[DanDan]

Interesting replies Ethan. My RT not defined thing is a bit pedantic. To be mathematically diffuse over a decent frequency range requires a large room. There is an actual size defined by ISO. The same ISO and others take Rev measurements at various room locations and do a simple mean average for each band. We have all seen the difficulties in measuring LF absorption in large commercial Rev chambers. This is because the rooms are generally not big enough, or averaged enough to give accurate RT figures. Of course this makes your data look even more extraordinary. I have always wondered why Fuzz does not deliver decent EDT graphs. One can often see a zero time octave band. I have usually put this down to the fact that the sound field is not actually diffuse at all. I believe Chris has taken all this averaging on board extensively in FM3. I would love to see your room tested with this or other working software. Dirac looked very seriously good to me on a few brief encounters. Perhaps you can use a demo of it, it is PC. I wonder is REW as good as it looks here, or is it actually applying fairly similar small numbers when uncertain or out of range. Do the Waterfalls tell the same story? Sorry, but it looks far far too good to be true! I would love to hear from REW's creator on this. Over what range of RT is REW accurate or even defined? Do forgive me if I am a bit sceptical. I have recently been trying to duplicate the remarkable Helmholtz experiment on page 229 of the MHOA.
It is not working out, I have built a similar device with no effect whatsoever on the room mode.
Best, DD

[JohnPM]

Originally Posted by DanDan

I wonder is REW as good as it looks here, or is it actually applying fairly similar small numbers when uncertain or out of range. Do the Waterfalls tell the same story? Sorry, but it looks far far too good to be true! I would love to hear from REW's creator on this. Over what range of RT is REW accurate or even defined?

REW doesn't guess, approximate, invent or otherwise fiddle the RT60 figures They are derived in the usual manner by applying linear fits to the Schroeder integral, REW uses an iterative procedure to estimate the best starting point for the integration, often called "Lundeby's Method" (from the paper by A. Lundeby, T. E. Vigran, H. Bietz, and M. Vorländer, “Uncertainties of Measurements in Room Acoustics,” Acustica, vol. 81, pp. 344–355 (1995)). The correlation coefficients for the linear fits can be viewed to get a feel for how reliable any particular octave or 1/3 octave band figure is. On the specific RT measure displayed, you can choose from EDT (line fit from 0 to -10dB), T20 (fit from -5 to -25dB), T30 (fit from -5 to -35dB)and what I have chosen to call "Topt", which is an "optimal" decay time estimate based on the slope of the Schroeder curve over a variable range chosen to yield the best linear fit. If the early decay time is much shorter than T30 (often the case in small rooms) the Topt measure uses a start point based on the intersection of the EDT and T30 lines, otherwise it uses -5dB. REW then tests every end point in 1dB steps to the end of the Schroeder curve and chooses the one which gives the best linear fit. I find that to give more reliable and repeatable results, particularly in small rooms where the early decay can go through a couple of slope changes before settling to something more linear.

[DanDan]

Hi John, thanks for joining in. I am kinda gobsmacked here. As another poster has said Ethan's graph is 'extraordinary'. I have never seen anything like it in my practice or in any textbook. I use Ethan's Traps and others in treating mostly small Control Rooms. I have never seen such a dramatic improvement. That may be partly because I have never gotten useful EDT information from FM2, although I find the Waterfalls very reliable and can happily work away with them. So I hope you and Ethan both will forgive me if I poke at this a little to be sure of it. Something 'extraordinary' has happened here and I am very excited about it.
RT60 etc.
In my experience FM2 does not deliver reliable EDT T20 T30 graphs in small rooms.
Ethan has described a very similar experience with ETF.
Bruel and Kjaer equipment (costing multiples in tens more) goes to a great deal of trouble to get over the uncertainties. It uses multiple sweeps. It has 'quality indicators' for the results, prompting one to go again when it is not happy with them.
ISO RT measurement techniques use averaging with multiple source and measurement locations.
Clearly it is not easy to get trustworthy decay readings in small rooms.
I suspect that the reason for this would be that RT60 is not defined in small rooms. The sound field is not mathematically diffuse over the frequency spectrum of interest.
Thus any formulae based on classical RT60 theory cannot be fully valid in small rooms, and results using these formula will be compromised.
John, from your description ( and I am no mathematician) I can tell that you have experienced similar flaky results in experimentation and you have found a way out. In my simplistic view, your method seems to be based on using several classical approaches simultaneously, plus a good deal of your own seasoning and special sauce.
To me right now, this looks like a major breakthrough. I applaud you for that.
I don't have a PC, but if this all turns up fully true I will buy one! Perhaps it is time for an Intel Mac, then I would be able to move up to FuzzMeasure 3 also.
In the meantime perhaps you would be so kind as to advise and provide me with a version of REW suited to my Mac G5 OS 10.4. I will send you a PM to get in private touch.
Ethan, could you please show us the Waterfall? I am getting a bit suspicious. This miracle room thing, the multi-instrumentalist thing, the cat thing........
.Are we talking Wicca here or have you been to the Crossroads?

DD

[soundslikejoe]

Originally Posted by DanDan

Clearly it is not easy to get trustworthy decay readings in small rooms.
I suspect that the reason for this would be that RT60 is not defined in small rooms. The sound field is not mathematically diffuse over the frequency spectrum of interest.
Thus any formulae based on classical RT60 theory cannot be fully valid in small rooms, and results using these formula will be compromised.

Are there any publications that define a minimum area volume for RT60 to be accurately used? If it's "not defined for small rooms," then what is the cutoff size for accuracy?


(Obviously in over my head but having fun and learning stuff.)

[DanDan]

The lowest frequency tested should not be lower than 125 x the cubic root of (180/Volume) Hz. So there!
British Standards say that for lab testing of partitions etc. the minimum size of a rev room is 100Cubic Metres.
For Absorption testing in a Rev room the minimum size is 200 Cubic Metres.
These Semi and Reverbrant Rooms are good for tests down to 100Hz only. To work below that frequency would need much larger rooms. Most Labs won't even attempt measurements below 100. I wouldn't sweat this 'not defined' etc. My real point there is that the Maths will not be applicable because it was written for large room conditions. Even in large spaces, the mathematical predictions are rarely close to the measured reality. I am suggesting that these weakness in the Maths may be the reason why several serious attempts at software to measure EDT are not very convincing, i.e. ETF and FuzzMeasure. (Note I don't have FM3) Even Bruel and Kjaer with their twenty times the price product, take extreme care to get useful results. It is looking like John has combined various elements of classical Maths with some innovative techniques of his own, to come up with a reliable way of measuring small room decay times over a useful spectrum. In that case we really do have a breakthrough here.
DD

[JohnPM]

The key to whether the results are meaningful is the shape of the Schroeder integral. In a large room with a well diffused soundfield the sound decays logarithmically, which means a plot of the level in dB against time is a straight line. The measures of RT60 are built on this assumption of a log decay, and they tell us the slope of that decay line in seconds per 60dB. In small rooms, however, the soundfield is often dominated by specular reflections and the decay no longer follows a log trend. Best-fit straight lines to the Schroeder curve can still be derived, of course, you can put a best fit line through any data, but the answer becomes of little value. A measure of how well the data is following the line is the correlation coefficient, REW flags as suspect values with a correlation coefficient worse than -0.99 - the values are shown in italics in the cursor field. Other software will doubtless treat that differently, and choose their own thresholds, hence the "0" values you mentioned.

The big increase in response uniformity in Ethan's results look to have come when a lot of diffusion was added to the room, which would make sense, but we'll have to wait for the video to see just how extensive the treatment was.

[Chris Kress]

I would really like to see a room that small actually work... and you have to admit that Ethan's diffusers sure look sexy.. Maybe some day, I'll pull down my -leftover- T-fusers and and spend the 2k$ to redo my rear wall.. I'll bet my clients would make better comments while sitting on the couch in front of them.. But wow.. 2k$ just because you suspect. This video had better be good

[Ethan Winer]

Originally Posted by DanDan

Hi John, thanks for joining in.

I second that emotion.

Over what range of RT is REW accurate or even defined?

The lower limit appears to be fixed at 80 Hz. Which makes sense.

Ethan, could you please show us the Waterfall?

Glad to, attached. One is with the room empty and the other is with all traps and diffusors in place. Forgive the poor appearance of the graphs - I reduced them to 16 colors to save forum bandwidth.

I am getting a bit suspicious. This miracle room thing, the multi-instrumentalist thing, the cat thing........
Are we talking Wicca here or have you been to the Crossroads?

Just good science.

--Ethan

[jwl]

Yeah, I always pegged Ethan as more the Druid type than the Wiccan type anyway....

[jwl]

See, I have proof! Ethan IS secretly a Druid!

[DanDan]

My failing eyesight or other faculties have once again let me down....
I totally failed to notice that the RT graphs began at 100Hz, I just assumed full range.
That totally changes everything.
The Waterfalls show the usual horrible suck outs and booms.
Nothwithstanding that, the rest of the spectrum looks great and is a fine example of what can be done. Furthermore the Hearing is Believing bit is a great move. I believe many audio situations can be usefully shown by recording and playing back. e.g. Microphone, Preamp, Speaker, comparisons.
Unfortunately the region below 100Hz is the one I am most interested in.
It is of course the most difficult to control or even to measure. It is an octave and a half or so of full range music, i.e. most music. Some would point out that this area is by far the most important, containing vast amounts of energy, and in the case of Rock, most of the Bass and Kick Drum sound.
I am disappointed that REW only goes down to 80Hz, but perhaps that is a better call than the others have made. Similarly I welcome the 'confidence' indicators which REW, B and K and others provide. I am sure Chris of FM is watching this and will respond with such features.
If we did consider this room in terms of full range and third octave RT it would have a rising RT towards LF, generally welcome. However, and sadly, it would very much fail the earlier mentioned BBC test of third octave RTs being within 10%
But so would my Control Room which sounds wonderful and works really well.
Oh well.
DD

[jwl]

Actually Dan, not to be pedantic, just want to be clear. The RT60 graphs start at 80Hz, because that's where REW appears to start with its RT60 measurements.

But I definitely agree, that under 100Hz is where most of the action is in small rooms.

[DanDan]

80 indeed, I presume that is the bottom of the 100Hz third octave band.
I am in communication with John about trying REW and an experiment we have going on at studiotips. That experiment features 34Hz!
From what he said above, I still think he may be way ahead in terms of getting useful Decay numbers, T opt is absolutely fine by me if it is repeatable. Maybe I can persuade him to go down to the 31.5 Hz band, with health warnings.
Best, DD

[JohnPM]

Originally Posted by DanDan

I am disappointed that REW only goes down to 80Hz

Ah, we have been talking at cross purposes. The business of making acoustic measurements is set out in an ISO standard, ISO 3382. That recommends analysis in the six 1-octave bands from 125Hz to 4kHz. Most measurement software, REW included, extend this to include an extra octave above and below so 62.5Hz to 8kHz. To extend the range to include the 31.25Hz octave would really go beyond the point where an RT60 figure tells you anything meaningful at all, as the lowest filter would then cover such a narrow range of frequencies that even in a large room the behaviour of relatively few isolated modes would dominate the result, giving Schroeder curves that bear little relation to straight lines. Many software packages, again REW included, offer analysis in 1/3rd octave bands, but it would then be typical to restrict the lowest frequency to the band centred on 80Hz to have some hope of spanning a sufficient frequency range to allow a meaningful result. I would personally use waterfall and spectral decay plots for everything below 200Hz.

[Ethan Winer]

Originally Posted by DanDan

I totally failed to notice that the RT graphs began at 100Hz, I just assumed full range ... It is of course the most difficult to control or even to measure.

Just to be clear, it's only the RT60 display that stops at 80 Hz. As John said, below that is useless as reverb time anyway, so instead we look at the waterfalls which in this case go down to 30 Hz.

--Ethan

[DanDan]

Claro, Amigos. However I do not at all agree that below 80 can set aside, due to measurement or other difficulties. Later ISO and other Standards use extended spectra e.g. 63Hz - 10K and such. They extended because they were getting wrong results.
I believe John is working on some innovative ways to look at the lower frequencies of interest, getting away from Classical definitions and formulae. What I would like to to see is 'how long a room rings on' in third octaves. If a particular band is dominated by a mode, I don't really mind, that is still good info and the Waterfall will indentify the mode. I believe the shape of the decay spectrum is much more indicative of a room's sound then either Freq. Resp. or Waterfalls. I look forward to John's new takes on this.

DD

[Ethan Winer]

I hear you Dan. I'm not an expert with the standards etc, but wouldn't this be mainly averaging the same data used in the waterfalls, but divided at the standard third-octave frequencies? More to the point, averaging mode decays into bands kinda hides what we care about. That is, a small room has extended decays only at certain frequencies. So it never seemed important to me to consider anything else.

--Ethan

[DanDan]

I think we are all in the same choir Ethan, but some different hymn sheets have been distributed by Mrs. Palin probably. Or, there are many ways to skin a cat......OOPS :-)

I think we all want a room that enables the bass instruments to be heard clearly and separately, without 'single note bass' or overall boominess or not enough overall bass.

We differ in only in how we like to view the measurements, and perhaps the final goal.

You like to identify and quantify individual modes, comb filtering and such, I note you also like to use little or no smoothing when viewing. I am guessing our goal is a flat line for frequency response and perhaps decay.
I go the other way. I want a skewed response, diminishing a HF. Old School. Ditto the Decay times. Perhaps because of this I rather see broad trends, the slope of the curve rather than the wiggles. I like third octaves for several reasons. I am very used to them from live work. They do have a physiological connection with how the ear works. I do believe they visually and otherwise come closer to describing what it sounds like.
They were designed for this work.
Of course I would turn off all smoothing when investigating a single mode, or a single path, to get forensic.
A single mode within the span of a third octave will of course give a very long reading, or it should, if the measurement system can handle the distorted non logarithmic decay. So I think we will end up with the same results.
We are both after the same thing, the cat!
DD

[Ethan Winer]

Originally Posted by DanDan

some different hymn sheets have been distributed by Mrs. Palin probably.

LOL, she just resigned, so we won't have her to kick around any more.

you also like to use little or no smoothing when viewing.

Yes, because smoothing and averaging etc just hides the detail. Give it to me straight doc, how much longer do I have to live? Don't sugar coat it!

--Ethan

[DanDan]

Indeed, scary to think she might have been VP.
That resignation speech was quite unhinged.
I hear you Ethan re the detail, but smoothing and averaging make it easier to see the slope, revealing a detail called the bigger picture. Similarly third octaves were developed for a quite a few reasons. They do correlate with audible phenomena much better than single tones. However I, and probably you, use both, smoothing for the big picture and no smoothing with zooming to get at areas of particular interest.
DD

[Ethan Winer]

Originally Posted by DanDan

third octaves were developed for a quite a few reasons. They do correlate with audible phenomena much better than single tones.

Yes, mostly, but not always. Maybe you've seen this short article:

Audibility of Narrow-Band EQ

I, and probably you, use both, smoothing for the big picture and no smoothing with zooming to get at areas of particular interest.

Exactly. I generally use third-octave smoothing above 500 Hz or so, but mostly because it's very difficult to see through the severe comb filtering that is always present but not necessarily audibly damaging. I recently calibrated the active crossover on the big speakers in my studio and I used third-octaves in REW to better see the big picture of lows level versus highs level. So on that I certainly agree with you. However, if I were measuring a loudspeaker in an anechoic chamber I would not use smoothing unless I were the manufacturer and wanted to hide the true response from potential customers.

--Ethan