Tailoring pink noise to root out troublesome freqs

[macc]

Hello all,

I’ve been using the RealTraps pink noise file and finding it very helpful. I was thinking about using pink noise filtered even more, in order to tailor it specifically to the region (a sort of broadish 80 – 140Hz dip) that I am having trouble with. My thinking says that using an SPL meter should then tell me where those frequencies specifically ‘live’ in my room, and help guide trap placement to reduce the problem.

Is that a fair way to go about it would you think? Has anyone tried this approach of tailoring pink noise for more specific problems?

Any thoughts appreciated, cheers!

[Ethan Winer]

Unless you don't have a computer, dedicated software is a far better way to analyze a room.

ETF, Windows, $150

FuzzMeasure, Mac, $150

Room EQ Wizard, Windows and Linux, Freeware

This article explains how I use ETF, but the principles apply to all such programs.

--Ethan

[macc]

Hi Ethan,

Perhaps I wasn't clear enough, sorry. I am aware of those bits of software, and have been analysing my room using REW and the ECM8000 you recommended to me a few weeks ago (thank you!).

It was a combination of the analysis results and your comments in the 'pink noise aids bass trap placement' article on your site that got me thinking. I mean 'some frequencies may end up more in one corner than another and vice versa - depending on the loudspeaker and subwoofer locations and other factors. So the goal is to find where bass energy accumulates the most, and put your bass traps there.'

REW tells me I have a dip in that area, so I thought one could use tailored pink noise to find which corner the problematic frequencies have 'ended up' in. Then treat accordingly, analyse the room using REW and see what improvements can be made. So, a sort of combined approach.

Am I just being dense in thinking this could be a useful method? It wouldn't be the first time

[Ethan Winer]

Originally Posted by macc

REW tells me I have a dip in that area, so I thought one could use tailored pink noise to find which corner the problematic frequencies have 'ended up' in.

Okay, got it now. Yes, that should be a good plan.

--Ethan

[macc]

Nice one Ethan - thanks a lot. After reading about a zillion of your posts and reading your site front to back to front, some of this is starting to sink in :D

It has to be said the improvements have been... well, staggering really. Thanks a lot.

[SaSi_SiDi]

Wouldn't it be the same if REW or ETF was used with the mike located in various places in the room? I would assume that each location would give a different waterfall chart depending on what is going wrong in each location.

[macc]

I suppose so, that could help. Seems a little long winded though, that is, I know that around the listening position, the region around X Hz is a problem. So playing the region around X Hz with pink noise and then walking about with an SPL meter will tell me rather quickly where I need to trap.

That's my thinking anyway...

[SaSi_SiDi]

I was wondering how can one filter out pink noise to only include a certain frequency range. Would a very steep filter introduce higher frequency artifacts?

Would a warbled tone have a similar effect?

[macc]

I wasn’t intending to use a ‘brickwall’ filter – just to roll off more of the top/bottom to gently bandlimit the noise to the desired region. A 24dB/Oct filter (or maybe heavy shelving) would be more than enough. As for higher frequency artifacts, I don’t really know what you’re talking about there. Certainly, using a very steep filter would cause a great change in the phase response, but being pink noise – and as such not ‘timing-critical’ – I can’t imagine why this would be a problem.


I imagine a warbled (ie pitch modulated, if I read you right) tone would be useful as well, but it wouldn’t be constant at all its consituent/reproduced frequencies as it moved about. You’d have to have the modulation very fast I suppose, but yeah, I reckon that could work.

[Ethan Winer]

Originally Posted by SaSi_SiDi

I was wondering how can one filter out pink noise to only include a certain frequency range. Would a very steep filter introduce higher frequency artifacts?

I can't see why. If you make a band-pass filter with an EQ plug-in by combining LP and HP filters, all the highs are reduced. Even with a very high Q, that affects only content around the cut-off frequency.

--Ethan

[SaSi_SiDi]

Originally Posted by Ethan Winer

I can't see why. If you make a band-pass filter with an EQ plug-in by combining LP and HP filters, all the highs are reduced. Even with a very high Q, that affects only content around the cut-off frequency.

--Ethan

I am refering to using a digital filter. I remember reading sometime ago that a steep cutoff digital high pass signal would generate some sort of ripple in the harmonics of the cutoff frequency. I will try to locate the site and article.

[jwl]

For the purposes of your test, just take the pink noise file you've been using and slap an EQ on it with both a highpass and lowpass filter on it. Tailor it to the frequency range you desire, and re-run your test with the newly-EQ'd file.

[macc]

Erm... thanks. What did you think I was going to do?

[Ethan Winer]

Originally Posted by SaSi_SiDi

I am refering to using a digital filter.

And far as I know, digital and analog filters having the same number of poles and Q will behave exactly the same. Yes, you can get ringing with very high Q filters, but I'm pretty sure the ringing is at or near the cut-off frequencies. Versus an octave or more away. But do look for your reference and post it if you can. I've been wrong once or twice before.

--Ethan

[macc]

Reporting back:

GREAT success. I did the test as described with the LP/HP on the pink noise. I went round the room watching the SPL meter, listening... I felt like I was in Ghostbusters I made marks on the walls (near the corners) where the meter really went up, and smaller marks where it was it was up but not so bad.

Redistributing the treatment I had (which was just sort of thrown in until I had time to do proper analysis) according to this method gave me an 8dB improvement in the targetted region! It also really really (!) smoothed things out throughout the entire spectrum. The room sounds soooo much better!

Maybe a more experienced person could do this without the added step of filtering the pink noise, but it doesn't take long and it has really helped me. Still more improvements to make, but for a tiny temporary room it sounds really really good

[Ethan Winer]

So your Professor John Frink experiment worked? Excellent!

[macc]

That looks disturbingly similar to your avatar Mr. Winer


(joke!!)

[Ethan Winer]

I admit I'm the biggest geek in the world. I only wish I were as rich as Bill Gates.

[jwl]

Originally Posted by macc

Erm... thanks. What did you think I was going to do?

Ah, sorry. I thought you were asking the best way to filter the pink noise file.

Glad to hear the test worked out for you.