Ethan wrote in an article (EQ, April 2004):

Standard real time room analysis using pink noise to measure the frequency response in third octave bands completely misses these peaks and nulls. When pink noise is analyzed in bands, the levels of all frequencies within each band are averaged together. Even measuring at 1/12th octave spacing is far too coarse to see the true room response. I have observed peaks and adjacent nulls less than 1/12th octave apart in many small rooms. So depending on what frequencies are measured versus at what frequencies the peaks and nulls occur where you place the measuring microphone, it's likely that a room will appear perfectly flat when in fact there exist many large aberrations that are completely hidden.

If I'm not mistaken the human ear can't distinguish deeps that are very narrow (that's why notch filters are used). Is the above maybe a tad picky?

Dino,

> If I'm not mistaken the human ear can't distinguish deeps that are very narrow (that's why notch filters are used). Is the above maybe a tad picky? <

Nope, not picky at all. If a particular listening location gives a null that's, say, 35 dB deep at 110 Hz, I absolutely guarantee it will be a serious problem for all music in the key of A.

--Ethan

So, is pink noise no use at all, even with different analysis techniques?

Dino,

> is pink noise no use at all <

Pink noise or sine waves can both be used. The issue is how finely the results are analyzed and displayed. For example, if you have a way to assess the measured result of pink noise at 1/24th octave resolution, that's fine enough.

I use the ETF program which tests using either a swept sine wave or MLS, which sounds pink noise. So either can work. It's all in how the data is analyzed.

--Ethan