Update: PC Interface
Working with the Copland PC interface was both intriguing and instructive. The display consists of a logarithmic graph of frequency (X) and amplitude (Y) overlaid with three sets (left and right channel) of curves: The first set represents room response, the second represents room correction applied and the third represents the net (corrected) response. A line through 0 db represents the target.
In my system, the display showed a room having very uneven response. Above 1 kHz, there were several broad excursions having peaks and troughs of ~6 db above and below center. These the DRC corrected easily, yielding a net response that was within +/- 2 db.
Below 1000 Hz, there were two severe but narrow dips, the first being 10 db at a center frequency of 170 HZ and the second being 15 db (!) at a center frequency of 65 Hz. Since Copland does not recommend filters with a gain above 6db, I spent some time experimenting with speaker placement to see whether I could ameliorate the problem before adding more compensation.
Based on my reading, the likely culprit was either floor reflections or reflections from the rear wall. Tipping the speakers back ~5 degrees from vertical helped slightly, Toeing the speakers in smoothed response above 1000 HZ but aggravated both of the lower modes. Toeing the speakers out had the opposite effect.
Moving the speakers further from the front wall reduced the mode at 65 HZ but made the mode at 170 Hz worse, suggesting that much of the room interaction was coming from the wall behind my listening position. Some basic math suggested that reflections at both frequencies would be out of phase by nearly 150 degrees at my speakers. Given the wavelengths involved, the speakers would need to have been 6 or more into the room before the effects were reduced significantly. I settled on a position slightly inboard and forward of the original location with ~ 10 degrees of toe-in.
I now had 2 narrow, 10-12 db modes to remove. Copland allows the user to define 5 filters of varying configurations and Quality (Q) factors. Beginning with the default (A1) curve, I created a pair of peak filters having higher than recommended Q and gain that were centered at each mode. Although the contours of the response curve were very similar between channels, the amplitude was not. Because adjustments are applied to both channels, I had to settle for a final curve in which the right and left channels straddled center.
The PC interface permits the user to try curves before committing them. It also allows the user to switch among each of the filters. That said, if a filter is not saved before switching, it is lost. Configuring the filters was very easy, and the results were very audible. At the end of the session, I preferred my custom settings to the default by a slight margin.
