TOPIC: PEQ Before Dirac Live?

It sometimes get's discussed whether it is useful to run PEQ before Dirac Live. The answer will likely depend on the situation but I've made a comparison for my own which I thought I'd share. This is in the context of my SHD providing a crossover between my subwoofer and main speakers at 110 Hz.

On 13 December I completed a 9-point set of Dirac Live measurements and generated a Dirac filter based on the default curve. Today I did the same but with PEQ applied to flatten the response below 55 Hz first. The measurement points won't have been absolutely identical on the two days but I did measure the microphone positions in both cases so they should be fairly close I think. Within 2-3 cm I'd guess.

The following measurements are all at the same main listening position with 1/24th octave smoothing applied, and for simplicity just for the right channel. The first graph shows the response with no filters at all in blue and with the PEQ filter in red. The idea of the PEQ filter was to get the response closer to where it should be for Dirac to then refine.



Generating Dirac Live filters for each and applying them, the resultant responses at the same location are as follows, with blue being without PEQ also used and red being with PEQ. The results are remarkably similar! The small level difference I think just being due to slight differences in where the 0 dB level was set in DL for each. I wish DL showed absolute levels to be able to prevent things like this.



This shows the amplitude response of the combined filtering in each case is very similar but I wondered if there might be bigger differences apparent in the phase responses of the filters? The graph below show the phase responses of the filters (not the in-room responses), with blue being pure Dirac Live, red the PEQ first approach, and just for comparison the phase response of the PEQ filter on its own. The results for PEQ first is again remarkably similar to having just run Dirac Live on its own.




The take-home for my own situation is that there doesn't seem to be any obvious advantage in generating and applying a PEQ filter ahead of making Dirac Live measurements and generating its filters. At least not in the low-bass region where I wondered if it might have a role.

One thing that occurs to me is that Dirac can easily apply almost infinite cut so cutting a peak with PEQ probably won't help it much. Boosting a notch with PEQ however could possibly help it with the very limited boost it can apply.

In your specific case there is a series of peaks and notches between 100Hz and 1kHz with an overall downward trend If I was to use a PEQ anywhere I would tend to apply a boost with low Q to either cancel this downward trend or even shift it to an upward trend. However, in your specific case the amount of ripple is likely within Dirac's ability to control.

Thank you for sharing that information @ultrasonic.

I found setting up DIRAC with one configuration using focused mic placement and one configuration with the wide setting has yielded a much more noticeable difference than the PEQ prior to DIRAC attempt I made. I did not bother to measure, and I'm glad to see that your results confirmed my subjective listening experience.

entripy wrote:
Yes, you're right that PEQ before Dirac Live could be used to increase the maximum boost applied to above the 10 dB limit of the latter. I'd question how wise this would be to do though. If Dirac Live had the option to adjust the maximum boost it uses I'd personally be inclined to reduce it.
entripy wrote:
The general downward trend of the post-Dirac response is the result of the default Dirac target curve and is I'd say desirable rather than something to look to 'correct'. It could easily be changed by adjusting the Dirac target curve should somebody wish to, rather than requiring any use of PEQ.

Also bear in mind that this is a single point measurement whereas Dirac Live works on the basis of a spatial average, and that where the listeners two ears may be will always have a least one getting a different response to a single point measurement.

I actually have more experience with corrections based on a single point measurement and know that I could make the bass response smoothly fit pretty much any smooth shape I choose at a single point. I must admit I do therefore struggle a little with seeing single-point Dirac results that don't look as good but I do appreciate that single-point EQ can give a false impression of what will be heard.

I've just updated the phase graph in my first post to remove time delay effects highlighted to me by @mdsimon2 in another thread. It doesn't affect the main conclusion but shows this aspect more clearly.