TOPIC:

I have been using Dirac for 2 years and I love the improved sound that I get with it.  I am not posting this to bash Dirac in any way because I am a big fan and it is making my music sound better than ever.  I am trying to get the best sound I can in a non-ideal location for audio (my computer room) without spending too much.  I have used Dirac with multiple bookshelf speakers with and without subs and it has improved the sound of all of them.  I am a tinkerer (and I am retired) so I have spent lot of time making changes, trying to learn and improve the sound.  I have experimented with PEQ settings before and after Dirac, Dirac without PEQ, REW generated PEQ settings, Dirac with vertical and horizontal mic, Dirac with multiple and single mic measurements, etc.  I have also experimented with house curves and read everything I can find about Dirac, room corrections, trying to learn how to get the best sound.  

I am finding that Dirac frequency response predictions (at least some of the time) do not match subsequent REW measurements with Dirac active.  Maybe I am doing something wrong to cause this.  This has been bugging me for a long time so I decided to do an experiment.

Here is my system:  Quad S2 speakers, Yamaha RX-770 receiver, DDRC-24 (firmware and software are current), UMIK-1

My experiment:  Use DDRC-24 PEQ to create an approximately flat response with Dirac inactive.  Document flat response with TrueRTA and REW.  Create Dirac filter with flat target curve and DDRC-24 PEQ settings active.  Document frequency response with Dirac filter and PEQ settings active.  Do this experiment with single measurements without moving the mic.  Compare measurements before and after Dirac.

Please see attached PDF with screen shots showing the experiment.  TrueRTA pink noise generator was used for TrueRTA frequency response measurements.

Can you explain or tell me what I am doing wrong?

Thanks!

I assume that you carried out a standard Dirac Live calibration with measurements at multiple points? If so then the important thing to realise is that the Dirac Live algorithms work with some sort of average of all individual measurements and is targeting a smooth response across the measurement volume.

It's not a good idea for sound quality but it is possible to run a Dirac Live calibration based just on a single measurement at the central measurement location. If you do so you'll find the corrected graph in Dirac will match a frequency sweep measurement in REW at the same location pretty well (I've tested this myself before).

I've never used TrueRTA but have used the RTA functionality in REW when I've wanted this.