Join Date: Sep 2010
Location: Huntington Beach, Ca
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 17 Post(s)
First, I'll suggest the concept of maximum phase doesn't exist, and that this might be what we call linear or zero phase. This is where the filter time aligns all frequencies: corrects for amplitude and all phase rotation (speaker and filter). Minimum phase is where the filter doesn't correct for phase, and in addition, adds it's own additional phase rotation as a result of amplitude corrections. The EQs that everyone is used to: tone controls, graphic and parametric EQ, they are based on IIR filters. One could assume that linear/zero phase filters would always be ideal, but not the case at low frequencies. In typical rooms that affect the bass amplitude response, there is a corresponding phase rotation (in addition to the rotation that may be found in the speaker). An appropriate IIR filter used to correct the amplitude error will also correct the phase error naturally. The phase rotation of the filter will offset or counteract the speaker/room phase error.
We provide the user several options and combinations, as well as the ability to define the filter sizes that are used by the automatic correction algorithm. Typical setting is for a combination of FIR linear/zero phase for full band, with any range of ms or points, combined with IIR for low frequency, again with user definable setting for frequency limits and number of filters, etc. Other choices for the FIR is the ability to use minimum phase, align all speakers' phase according to Left/Right, or use of FIR or IIR alone. Align according to L/R is done when one wants to keep the sonic signature of their L/R loudspeaker and impose this on the remaining speakers (some like the sonic signature of phase rotation, with consistent rotation, speaker to speaker, for ideal for imaging).
To summarize, there are default settings, which many rely on, then if one wants to exert more control over any variable, the ability to do so is there. For comparison, one could set the Trinnov variables to match another component under test, thereby providing one with a starting point for evaluating the relative merits of the automated algorithms, which can be a very different animal all together. What the algorithm looks for in building the filters is key to the results. Ours has a number of proprietary routines that I'd love to share, but then we'd be giving away the recipe, so to speak.
3D Audio Consultant