Parametric Equalization

Parametric EQ is a method for adjusting the frequency response.  It uses a number of individual filters that can be centered at a given frequency and then apply boost or cut at that frequency.  Typically, Parametric EQ is used in the production of content to apply the preferences of the content creator to the octave-to-octave balance of the sound. Unfortunately, it has also been used in trying to correct room acoustical problems.  It is not a method that is well-suited to fix such problems for many reasons including: (1) the types of filters used can cause audible artifacts and (2) there are not enough filter bands to address the finer peaks and dips that show up in a room.  

MultEQ uses a different kind of filter that is not based on Parametric EQ.  These filters have hundreds or thousands of coefficients that can be individually manipulated to make very fine corrections in the frequency response, but also in the time response of the signal that is affected by reflections from surfaces in the room.  Parametric EQ methods can't address these problems at all.

Can you explain a little about how MultEQ deals with the time domain?

Is it in the measurement or the correction you apply?

MultEQ measurements are taken in the time domain.  That means that the test signal is captured after playing from each speaker and is analyzed to see how it evolves in time.  The first burst comes from the direct path, then from various room reflections.  This gives MultEQ a pattern of how sound from that speaker was influenced by the room and nearby surfaces.  Methods that use pink noise (or similar) do not have the ability to see the time domain and only look at the frequency response.

Once all the measurements have been taken, MultEQ performs an analysis of all the time domain patterns and groups them according the similarity of the problems it finds.  Then it creates a representative time domain response for each group.  These representative time domain responses are then combined to give a single time domain representation of the speaker behavior in the room.  The MultEQ filter is then created by inverting that final time domain response.  A filter is created for each measured speaker and sub and is then applied to the signal in the time domain via a process called convolution.

Chris, I'm experimenting just for fun with modifying a system's response with a greater degree of modification than Pro provides.  I tried an experiment utilizing first a good parametric equalizer, then doing a consumer MultEQxt calibration just to see what would happen.  The Audyssey on/off A/B comparison is very similar, although MultEQxt did improve the splice of one of the LCRs and subs.  In verification measurements, I can easily see those improvements (and the mid-range correction). I can't detect any audible artifacts, and I achieved the sonic objective.  Any reason why this can't or shouldn't be done if the situation is somewhat unusual?