Site Links
Howdy, Stranger!
It looks like you're new here. If you want to get involved, click one of these buttons!
Quick Links
Categories
Who's Online (4)
- DrewsBrews 1:54AM
- jholtz 1:54AM
- rjj45 1:54AM
- Tom_S 1:54AM
Room EQ such as Dirac
I’ve only just caught up and read the Dirac white paper.
I now understand that if I’m measuring my system in an un-gated manner, as I often do to see the bass/room modes, then the slightly downward sloping response is the goal, as the early and late reflections from the high frequencies are more easily absorbed and reduced in overall level. To shoot for a non-tilted response would be to tune the system too hot at the top end (as a gated measurement would show.)
I also understand a key aspect of the Dirac method is multiple measurements in the listening window to get a better picture of the room reflections, as a single measurement may contain an extreme peak or null not found elsewhere in the listening window.
What I want to better understand is this “mixed-phase correction.” The white paper emphasizes the analysis of the impulse response at each measurement location, but is this not just a simpler way to see which reflections are common between all measurement locations? After all, isn’t the impulse response just the time-domain representation of the frequency-domain response curve? So isn’t all that the Dirac program is doing is mathematically compare the impulse responses from all the different measurements to see which frequency anomalies exist at all locations and can then be EQ-ed in the frequency domain? Couldn’t this be done manually?
Cheers,
Comments
Mixed phase means they use FIR above Schroeder frequency and IIR for frequencies below. To do FIR down to 20Hz would need just too much compute.
Yes it can be done manually with a variety of methods. Accurate, Audiolense, REW and others can do it, but are not automated/wizard driven. SONOS Trueplay, on the other hand adopts the Apple approach- just wave your phone around the room and it will do automagically do it for the general consumer. Audiolense has the UI of something made in the Windows XP era, and something an engineer could use, REW is better (just)
AFAIK They were one of the first to automate the steps into Windows wizard type step-wise procedure for taking measurements for either 3 measurements around your listening position; or 9 measurements sparsely around the listening roll. Around the same time miniDSP collaborated with them to run in on their DSPs, so it would free you from PC. (I liked this because it would be like an pre-amp/appliance and wouldn’t confuse my wife)
And then went on courting manufacturer’s to license their idea. Eg. Car audio, headphone manufacturers
The founder recently left for Harman so it’s probably a signal that their tech is pretty stable.
I first came across it after reading a review in 2013, so it is a bit long in the tooth, and perhaps didn’t take off that much as it should have. Sometime technology needs aggressive marketing, and IMHO they are one of the ‘good guys’ (merciless honest technologist) . They even had an app that turned the crap gen 1 Apple wired earbuds into something that sounded quite good!
People who don’t like it have either used a free solution and don’t feel that their fee is reasonable, not necessarily because the Technology or sonics is bad. Also, sometimes people turn it off because they prefer how the room affects the sound.
Believe it or not, precise imaging around your listening positioning can get a bit stale, like how people like to rotate DACs or amps, whether they are any different (real or perceived)
I don't use any room correction. I don't need it.
InDIYana Event Website
If, as we do as DIY builders, voice our Xovers in our room, the room is captured (or at least somewhat addressed).
Yup
There will always be room modes in the bass frequencies. That is why we use multiple subs and move them around.
I like to play with the squiggles.
I have crazy room modes between about 80-200 Hz like most people’s rooms.
I’m not sure I’ve run across a compelling argument not to use my A/V receiver’s GEQ to try and make them better. The Dirac methodology seems like a sound method that I could begin to replicate (to a much lesser extent) manually.
I thought I read that any filter is going to introduce phase distortion, and having studied ME rather than EE the whole FIR vs. IIR is still a bit beyond me for now. But aren’t GEQs in A/V receivers minimum phase?
Cheers,
FIR allows frequency adjustments whilst preserving phase. Or to put it another way, you can adjust phase independently of frequency, this is why it doesn’t introduce those weird special effects, to put it politely, of previous room correction technologies, (including graphic or parametric EQ)
Here is one of my favourite speakers the Statements II in a 20x30x12’ room.
Measured not at the listening position, as is the purpose of Dirac, but at a “standard” 1m to see how it might compare to spec:
Here it is at the listening position, using 9 points:
To be clear, room correction doesn’t fix an acoustically bad room. The ‘best’ rooms are not made with aluminium and glass like your bathroom/shower. but nor are they an anechoic chamber.
But it can certainly fix the sound in rooms that you already have…
... and helps offset the 'textbook diffusers' behind the speakers as well.
Very "reflective" surfaces . . .
Thank you for the responses.
I’m not very well read on this topic and I’m genuinely curious.
Is it correct to say that any parametric or graphic equalizer is going to alter the phase response of the speaker at the filter’s center frequency? This is the same thing as introducing group delay at the filter’s center frequency, much like a crossover filter can? So this is the argument against their use, right?!
Thus the benefits of FIR and IIR are constant group delay as I understand it.
I’m also curious, do parametric or graphic equalizers typically introduce latency into the system? From what I’ve read, IIR and FIR filters do, correct? So it would be very difficult to use a FIR or IIR correction with an A/V system where matching the audio and visual outputs in time is critical.
So if I’m optimizing for an A/V experience, I should use GEQ sparingly and preferably only at lower frequencies where group delay distortion is less evident/objectionable?
Cheers,
My limited 2 cents.
As with passive notch contours; use sparingly, address peaks more so than dips. The issue of phase shift is more problematic (IMO) when a filter is placed on one driver through a Xover range biz then you are getting two drivers playing with relative phase differences with additive and cancellations - (crossovers and contours). Placing a filter before the system will still alter phase over the range.
I would not use the word "fix", but assist in tonal balance through mid-bass range <800Hz. EQ is not a fix-all solution, it won't correct directivity / power response problems, nor will it correct overly resonant RT60 decay problems like the room in your photo. However a couple bands of PEQ through the mid-bass can be beneficial for any system in any room,
Passive filters, active op-amp filters, graphic EQ, PEQ are all examples of IIR filters. They don't inherently add delay / latency, however frequency response and phase are directly tied together so you can't adjust one without affecting the other. IIR EQ is no more damaging than correcting frequency response in a crossover, its more of an amplitude compensation for the damage already done by reflective surfaces in your room.
DSP does add some delay, for hardware DSP typically only 2ms or so so, not enough to care about as far as A-V sync goes. For software DSP may be more like 20-50ms. Many AVRs run the signal through a DSP whether you want it or not.
FIR filters allow for independent control of amplitude and phase. This gets a lot more complicated, and there is a catch in adding delay to the system for low frequency corrections, as the most useful application of FIR filters would be to correct for group delay at low frequencies, however this requires "many taps" for the long delays which definitely will cause A-V sync problems. Not an issue for music playback, but any A-V or live/recording application is problematic.
For the "gateway to EQ" I would stick with IIR filters, PEQ stuff, and I recommend looking at measuring with "moving mic method" in REW for room measurement for EQ as a simple and easy way to get immediate results, before moving on to more complex methods. It's important that EQ is applied to a spatial average of measurements around the listening area, rather than to try to adjust for a single point is space from an individual measurement. Moving mic method uses RTA averaging to take many simultaneous measurements as you wave your mic around the listening area, and average them on the fly.
I had a treadmill on the FAR right of that photo. Just for laughs, I placed the mic there and follow Dirac's instructions of taking 9 measurements around where my head would be if on the treadmill.
Then when you stood on the treadmill and went for a walk/run, it sounded like the sound what coming right in front of you. not to the far left.
Pretty amazing...
Impulse response convolution can do some wacky and amazing stuff, but I'm not sure I'd call it "EQ" at that point.
Indeed…
https://www.celestionplus.com/
Deleted wrong thread