@Matthew J Poes frequencies below 200 Hz is typically the most problematic. Audyssey has done a very good job flattening out the response in my room. Have read a lot of good things about DIRAC. Trinov is another one that comes up a lot. Based upon my experiences w/MCACC, YPAO, and Audyssey I have gotten better results w/Audyssey. Thanks for the info.
Cheers,
Phil
Absolutely, but I've actually not had as good results from Audyssey as I have from manual approaches. DIRAC does a very good job and I find it to work well for that.
All of these systems work under different principles and with different assumptions for the algorithm. Some are better than others, but many of the mainstream, I think, get it wrong.
Audyssey's recent switch to an app that allows manual setting of the room curve and viewing of the results was a nice addition. I think this helps for when the algorithm mis-judges the EQ curve in room. Dirac has always allowed this to be manually set, but I think the concept is poorly understood, so people misuse it a lot. MCACC, YPAO, etc. are all using inaccurate room curves and nearly always get the results wrong. As far as I am concerned, they are useless.
One of the things that has always made me nervous about DIRAC, Audyssey, and even Trinnov, is how they handle the highest frequencies and what correction is applied. The approach to measurement that we use for in-room correction involves aiming a microphone at the ceiling and then measuring at different locations. We correct the response against a reference mic. Hopefully one that is corrected against a 0-degree measurement. What isn't addressed is that the change in mic position and the fact that it's pointed at the ceiling creates some uncorrectable errors in the response above 10khz. If the software attempts to compensate, the in-room response of the speaker will be wrong. I did a test of this using a very accurate microphone and careful measurements and I found that the microphones that come with Audyssey can't be corrected above 10khz to be accurate. The response didn't just roll-off that high, it had some problems that appeared to be inherent in the capsule or mic. Even the Dayton Audio UMM-6 with Cross-spectrum correction was not precisely correct in the 90 degree position, it's response was still more rolled off above 10khz than at 0 degree, corrected against a 1/2" refernce mic. To further complicate things, the changing in height changed the directivity of the mic relative to the measurement position at those frequencies causing changing errors. In other words, the angle relative to 0 degrees changes when the mic's position is changed, but the correction is only correct at 90 degrees. Below about 10-15khz this error is so minor as to not matter. Above that point the error becomes increasingly bad as the mic becomes more directional. When I looked to see if Audyssey and Dirac attempted to correct for this error, I found the answer was YES! It was! That's a problem because that was nothing more than a measurement error.
Then we get to a general problem with automatic room correction. It can't know the directivity of the speaker and a speaker's natural in-room response shape changes with directivity. The algorithm has to guess. Here again, YPAO, MCACC, etc. all get this completely wrong. I have to question if they even know what they are doing. Audyssey seems to have put some effort into this as it seems to adjust its correction against the natural response of the speaker, but it too get's it wrong when the speakers own off-axis response is less than stellar. Audyssey also incorrectly corrects very wide and very narrow dispersion speakers. Actually, I don't know if it gets very narrow wrong, but given its behavior with very wide, I believe it would do the same thing. It tries to apply a generic general room response shape to the response regardless of what it is, only adjusting things at the extremes based on the speakers natural response (it doesn't attempt to extend the highs or lows much),and this general shape is only correct for typical speakers. It is not correct for very wide or very narrow dispersion speakers and the magnitude of this error depends on the directivity. Further, it seems to make inaccurate correction judgments when the room has a ton of absorption. If it is extremely dead, and some people have a very dead room, the curve is incorrect.
Dirac makes the same errors but I found it easily addressable. I know how a room curve should look for various speaker directivities and decay rates, but I don't expect that to be normal (and with Audyssey, its not that addressable either).
This is why I am not a huge fan. Dirac is currently the only one I've used that I still recommend. I find its errors at mid/high frequencies to be minor if at all, it gets it right often, and I find it's ability to handle low frequencies to be about as good as the manual EQ approach, assuming the speakers are all setup correctly in the first place.
