Dirac and room correction software

[thumb5]

(repost from Wam Devialet thread, post 1382, 19 Jan 2014)

I've used the free trial version of the Dirac Room Correction Suite on my MacBook Pro, with the UMIK-1 USB calibrated measurement microphone available via their web shop. Just set it all up yesterday so it's early days, but I think I've already found out some useful things.

First of all, the calibration software does seem to work quite nicely using Devialet AIR as an output device, which was a pleasant surprise. I found that the target device buffer in AIR settings needed to be dropped to a low value (e.g. 50 ms) when doing the room calibration, otherwise the software reports an error during the analysis (I guess this is because the end-to-end latency is too high for it to correct).

Secondly, the software is quite easy to use once you've been through it a couple of times. The only part I found a little unnerving was setting the output level, which according to the software results in a much higher level than I'd normally listen to. I ended up using a compromise between what the software would accept and what I felt comfortable with.

Once the measurements are done, the software can build a filter to correct the room response: the recommended target response has a very gentle bass lift/treble cut slope. Alternatively you can tailor it to your taste, which I haven't tried yet. The filters are used by the audio processor part of the program, which runs separately and presents an output device to iTunes (or presumably whatever other streaming software you use), then sends the filtered output to AIR. The audio processor can switch in real time between one of four loaded filters, or no filtering.

In my tests so far, the filter has noticeably tightened up and evened out the bass. Other effects are much more subtle but I think I can hear improvements in imaging and ambience. I probably need to spend more time with the software to decide whether it's going to stay after the trial finishes. (Especially bearing in mind that the filters only work when streaming from my Mac, which is not how I normally listen to the 170.)

For your entertainment, here's the measured response for my room:



The blue lines are the left and right channel response (in each case averaged over 9 measurement positions). I was quite surprised by this - it doesn't look too bad to my untrained eye, apart from the modes at 30 and 50-ish Hz which correspond to the length and width of my listening room. This is without my sub running, just to get a feel for things. The red line is the recommended target response which is what I used as the basis of a filter.

---

(repost from Wam Devialet thread, posts 1429, 1432, 1434, and 1438, 25 Jan 2014)

I've just played around with FuzzMeasure to get some frequency response graphs for my room before and after using the Dirac Research room correction. Here's the result:



The two blue traces are left/right channel before correction, and the green traces are after.

These measurements are at a single point (in space) whereas the Dirac room correction takes samples at nine points and builds the filter based on the average response. I suppose this explains why there are some uncorrected peaks and troughs in the measured single-point response.

The generally lower magnitude with the room correction enabled is due to the DSP gain in the Dirac Audio Processor (-8dB by default), I believe.

In general the Dirac correction seems to have done a good job in flattening out the overall frequency response although there are still some trouble spots at (multiples of?) 75 Hz, particularly on the right channel. I should have a think about what could be causing those.

There's no doubt that the bass sounds tighter and more even with the room correction in place. Beyond that, the effect seems fairly subtle but if pushed I could convince myself that there's marginally better imaging and openness (not that either was lacking beforehand). So from a sound quality point of view, all good so far. Set against that is the inconvenience of having to use a computer as a source and to run the filter. I can see me doing this when I have time to play around and be a bit geeky -- and there's plenty of scope for that -- but it's not for everyday listening.

There's still a lot of air and openness. I will get round to playing with the target filter response at some point, but up to now I've focused on characterising the response of the room as best I can. Complicated by the fact that I tend to listen in two different positions -- seeing the difference between those is interesting. It's really handy that Dirac lets you set up multiple filters and switch between them (or no filter) in real time.

What I'm hoping is that when (if?) Devialet implement room correction in the amplifier itself, I know roughly what sort of difference it might make and have some idea of what correction needs to be done. This would (presumably) work with all sources and wouldn't depend on having a computer running, so would be very convenient once configured.