Testing room/speaker acoustics
I have invested many hours this weekend measuring various speakers in my 4 x 4m (approx.) listening room/study at home. I could not believe the results. I tried various (budget) microphones (such as the ECM8000) and finally used my master reference B&K 4133. I still didn't believe the results.
After about ten hours (wasted) I eventually discovered five factors that may be of interest to anyone who is contemplating measuring room acoustics and/or planning to use Digital Room Correction ...
1. The selection of measuing microphone is absolutely critical. I have measured three of these cheap uncalibrated mics. As (bad) luck would have it, the one I had at home - my No.3 ECM8000 - is up by a couple of dB's relative to my calibrated B&K reference 4133 at around 10kHz, and down by a couple at 20kHz, whereas my No. 1 ECM8000 (at work) has a slow roll-off starting at 10kHz. That's pretty good for budget microphones but do remember that unless you have a known calibrated mic to compare against you really are up a gum tree.
2. Don't assume (as I did) that CoolEdit/Adobe Audition generates pink noise with an absolutely flat spectrum. It doesn't. So any CD burned from this will exhibit the HF droop or whatever which will corrupt the accuracy of the room measurements.
3. Don't assume (as I did) that PC sound cards (and especially laptop sound cards) are flat with frequency: they are not.
4. Don't assume (as I did) that external USB sound cards are flat. Some are much flatter than others as I found out. Use 48kHz not 44.1kHz sampling rate.
5. Don't assume (as I did) that the CD player has a flat frequency response. Ditto the amplifier.
Taken altogether, the errors were far more than acceptable: an over-read of about 2dB at 10kHz followed by an under-read of -6dB at 20kHz and various problems below 200Hz. Insignificant in the course of normal music but completely unacceptable for reference measurements.
Recommendation: check the spectral content of the pink noise (loopback test, output to input) before you start. Use a known, calibrated mic. Yes, expensive.
Most of all, if your ears tell you it sounds OK but your test equipments doesn't, refuse to believe your test equipment and find out why before proceeding, however long that takes.
Last edited by A.S.; 09-05-2006 at 12:11 PM.
Alan A. Shaw
Harbeth Audio UK