Post Bristol 2015 thoughts - 5: room correction software, overview
NOTE: THIS POST (and a few relevant following posts) HAS BEEN CLONED FROM THE ABOVE THREAD AND MERGED INTO THIS THREAD>>>
We used the Dirac system (in playouts from my Vista laptop but never on CD direct to the amp) on Friday and Saturday but not at all on Sunday, when a Windows issue concerning a USB plugged into the laptop without consent then prevented it from seeing the main external music drive and the Dirac software from loading. So if you listened on Sunday, both on playout from the laptop and from CD, the sound was direct, unprocessed into the amplifier and on to the speakers.
I made a quick room measurement on Thursday night after we had set-up the room and furniture, and used the 'auditorium' setting, intended to give the best possible sound to a nine seated listeners in the middle of the room in a 2, 4, 3 seat arrangement rather than optimised for the more typical one or two users on a couch at the sweet spot at home. If time had permitted, it would have been interesting to try both modes.
Moving the microphone around the listening area and making a technical measurement at those nine points in space took about 20 minutes, and the Dirac calculation of average sonic behaviour across the seats and the subsequent calculation of a best-fit room correction filter just a minute or two. At that point, the Dirac system had made a suggestion for what it thought would give the flattest response (note: it cannot know about such esoteric concepts as 'best sounding') leaving the user to accept without input the suggestion or able to tweak the solution by boosting/cutting/shaping and levelling individual frequency bands at will. There was, in effect, an infinite override, down to fractions of an octave. That infinite flexibility is both an advantage and a great danger.
The basic low-end behaviour of the SHL5+ in the room was, as Sunday vistors would have heard (without DSP) surprisingly good. How much of that was due to the slightly dryer, tighter bass tuning of the SHL5+ relative to the previous models, how much due to the room absorber panels behind the speakers (more on that leter), and how much due to good luck we'll never know. We just plonked the speakers down where we though they looked OK in the surroundings and ticked that off the list. The DSP system was certainly capable (on screen) of perfectly smoothing the inevitable low end speaker-room lumps and bumps, not that is was a night and day experience: humans are remarkably tolerant of the bass effect of real speakers in real rooms. The issue with the Dirac which need much more study is not at the low end of the audio spectrum, it's because the Dirac is a full-range, wide band system, as capable of adjusting the signal passing through it at 50Hz or 20,000Hz with equal impact.
Once the Dirac system has 'swept' the room at the various measuring points, it presents the user with what it measured, averaged, and a suggested correction curve across the entire 20Hz to 20kHz audio spectrum. That curve can be saved. The user is then encouraged to slightly or significantly re-shape that target curve, and to save those as alternative room correction strategies. In theory, it would be possible to have one adjustment curve for every CD/LP or even for individual tracks, giving the listener precisely the sound he wanted, not only in the bass but at any frequency across the audio band: the ultimate graphic equaliser for both room, recording and taste. A tool of that power in the hands of an inexperienced user (that is, me) can be very dangerous indeed, and I found in the hour of so that I played with drawing (in the GUI) various slightly alternative overall target curves that fractional adjustments here or there of half a dB or so had a very significant effect on the subjective mid/top presentation. Yes, the response could be made ruler flat (or not, as desired) but that did not correlate with the best sounding (to my ears).
My impression is that restricted to the low frequencies only, say below 200Hz, the Dirac system could do a fine job of getting the best out of the speaker/room interaction that applies to all speakers in all real world rooms. However, extreme care and lots of accumulated personal experimentation is needed to confidently apply the DSP power in the midband, presence and top where unlike in the bass, the software tended towards wanting to boost the speaker output. If you do go for a room correction system then my latest thinking is that the AntiMode system with its limit of action to 500Hz or so is a much safer bet for the inexperienced user presented with an infinite possibility to screw-up a perfectly good sounding mid/top balance. The Dirac system is a fabulous tool, but it is dumb, and some weeks would be needed to learn how to tame its inherent enthusiasm to attack and smooth out every single lump and bump in the overall frequency response of the speakers in any given room when sonically, many should be left well alone, uncorrected.
The experience as a novice of setting up and applying the Dirac system has the same feeling as giving the keys of an Aston Martin to a 17 year old who has just passed his driving test and asking him to drive it with care. The inevitable over-reaction to the power and potential is guaranteed.
Alan A. Shaw
Harbeth Audio UK