Help needed interpreting measurement results

[PitterWaffle]

I finally got around to doing a measurement with the miniDSP UMIK-1 calibrated USB mic last november. I followed the steps described in the original manufacturer's website to the letter. And this is what I got (see attached REW ".mdat" file for more details).
The mic position is exactly where my head would be when mixing.

This doesn't look that great (those periodic spikes are not what I was expecting). It almost looks like there is a spike every 500Hz or so. (room mode?)

I do have access to a fully equipped woodworking shop so I could easily build anything I need to treat this room. What I'm not sure is WHAT to do with it. Do I need diffusers? Absorbers?

Any help you could give me to interpret these results (or get better measurements) and eventually get a smoother room will be much appreciated.

 

[John Mulcahy]

You need to use single sweeps when the input and output device are different, some of the variation you are seeing is due to timing misalignment between the multiple sweeps. A single 256k sweep is fine.

 

[Wayne A. Pflughaupt]

And, waterfall graphs are only relevant up to about 300 Hz.

Regards,
Wayne

 

[PitterWaffle]

Thanks everyone for the prompt reply.

You need to use single sweeps when the input and output device are different, some of the variation you are seeing is due to timing misalignment between the multiple sweeps. A single 256k sweep is fine.

OK, will do that. And should I be using one or two speakers when playing the sweep?

And, waterfall graphs are only relevant up to about 300 Hz.

I didn't know that, thanks. Is there another graph (using this same measurement technique) that will accurately display the lower frequencies?

 

[Matthew J Poes]

Thanks everyone for the prompt reply.


OK, will do that. And should I be using one or two speakers when playing the sweep?


I didn't know that, thanks. Is there another graph (using this same measurement technique) that will accurately display the lower frequencies?

He means it’s only relevant at low frequencies below 300hz. What he is talking about is that the point at which Decay is meaningful is the point at which the room is heavily contributing to the response which only happens between the modal region and the transition region. Anything in the stochastic region would only have meaning under anechoic conditions. Further, since speakers are minimum phase devices the ringing is just a result of amplitude variation. Not all that interesting.

 

[Wayne A. Pflughaupt]

As Matthew said, the waterfall graph is the one you want for low frequency decay.

Regarding the graph itself, 35 dB is a good lower setting, as most rooms don’t have noise floors quieter than that (unless perhaps you’re measuring something like a heavily-soundproofed recording studio).

For the time window, I think you’re about right at 400. If you’re decaying from 75-85 dB down to the noise floor in 350-400 ms, you’re doing pretty good, In my estimation. Perhaps Matthew can correct or clarify – I’m not much into acoustics measurements myself.

Regads,
Wayne

 

[PitterWaffle]

So I went back and re-did the measurement this time with only one 256K sweep like @John Mulcahy suggested. While I was there I got curious and re-did the same exact measurement with no params changed other than the sweep length. To my surprise what I found was that the longer the sweep the shorter the decay time in the waterfall graph while the top of the graph looked almost the same.

128k


256k


512k


1M

 

[Matthew J Poes]

longer sweeps reduce the noise floor. If it looks like decay time has shortened it is likely just a reduction in noise. Actual decay is not likely to have changed from that. If anything, there is an argument for longer measurements to more accurately energize the room and capture the decay information. For example the sweep vs chirp method.

In any case, anywhere the measurement isn't flat in the response is an area where the response could be improved, and the undulations in the response lead to peaks in the decay as well. Some EQ can help as would some acoustic treatment.

To see the benefit, go into the EQ tab, create an appropriate EQ target curve, generate the EQ filters, and look at the change in the waterfall. The prediction will be close to what you would see in real measurements.

See the article I just posted on the topic:
https://www.avnirvana.com/threads/eq-does-improves-bass-decay.2506/

 

[PitterWaffle]

@Matthew J Poes interesting read, thanks for the link.

 

[Matthew J Poes]

@Matthew J Poes interesting read, thanks for the link.

Thanks!

I looked at your plots a little more carefully. I didn't realize how high in frequency they went. So your bass is actually pretty flat and smooth below about 150hz. A little EQ might improve things a little, but I could also see someone liking that shape. It's at least smooth.

Above that point is the transition zone of the room and where rooms tend to just totally destroy your response. That is true of many rooms, not unique to yours. EQ can only do so much to fix that. Special EQ's like DIRAC are more capable of addressing this. For the most part, this is where acoustic treatment is most helpful.

 

[PitterWaffle]

Matthew, I really appreciate you taking the time to look into this.

For the most part, this is where acoustic treatment is most helpful.

I didn't know about DIRAC before reading your post so I did some research. I'm afraid DIRAC is a little out of my budget, but building a bunch of quadratic diffusers is (plus it will give me some much needed shop time now that the winter is giving up). Looks like something targeting 400Hz and up might be a good choice.

 

[Matthew J Poes]

Matthew, I really appreciate you taking the time to look into this.



I didn't know about DIRAC before reading your post so I did some research. I'm afraid DIRAC is a little out of my budget, but building a bunch of quadratic diffusers is (plus it will give me some much needed shop time now that the winter is giving up). Looks like something targeting 400Hz and up might be a good choice.

You can try diffusers. I was more thinking absorption. Diffusers that operate down to 400hz are usually fairly deep. Maybe a mix? Some diffusion and some absorption.

Dirac is expensive but at the moment its a fairly unique product on the market. I normally don't promote buying such expensive accessories, but this one happens to work so well that I consider it a worthy upgrade. It has an improvement on sound that is on par with other great components in a system. The effect is far more obvious than what you might get from an amp or processor and more on par with better speakers or an improved room (because it does improve the room).

 

[PitterWaffle]

Hmmm they have a free trial. I think I'll give it a go. Maybe the investment is worth it in the long run.

 

[Matthew J Poes]

Hmmm they have a free trial. I think I'll give it a go. Maybe the investment is worth it in the long run.

Yeah it can't hurt to try it. If you don't find the improvement beneficial, no harm done. If you aren't getting a much improved response that is disturbingly flat, let me know. Some people have had trouble getting the computer software working right. Can take a little tweaking.

https://www.minidsp.com/products/dirac-series/ddrc-24
Our friends at MiniDSP also sell products with Dirac. Might be an option. I am not sure what your setup is, but they have a variety of options.

I've always been a fan of this particular product, though it is quite a bit more expensive:
https://www.minidsp.com/dirac-series/ddrc-22da
The software you would be testing can handle up to 24 bit, 192khz. Many of the Dirac external processors are limited to 24bit and 48khz, but this particular one can do up to 96khz. If you don't listen to anything beyond redbook or don't care about the higher sampling rate, this is a non-issue.

Flavio (@Flak) hangs around here sometimes and can chime in if need be. He can be very helpful.

I happening to be listening to Queen right now with Dirac correction.