Quantcast
Room correction software (DSP) - optimising results for small rooms - Gearspace.com
The No.1 Website for Pro Audio
Room correction software (DSP) - optimising results for small rooms
Old 5th September 2018
  #1
Gear Addict
 
🎧 5 years
Room correction software (DSP) - optimising results for small rooms

OK, so I’m a fan of room correction software. My weapon of choice is Dirac Live (I’ve also used Sonarworks) and I achieve good results using it in my small home studio. In fact, I firmly believe that DSP software is essential for helping to control low frequencies in any small room (obviously in addition to acoustic treatment)… and perhaps even in some large control rooms too, not that anyone would admit to it But that’s a discussion for another day.

What I’d like to talk about here is how to optimise use of DSP for a typical ‘small control room’ set up, because the instructions that come bundled with Dirac and Sonarworks are - in my opinion - badly suited to such environments.

For instance: the recommended measurement mic positions (as suggested by Dirac and Sonarworks) are obviously designed for listening rooms, in which there is minimal furniture and other obstacles between the speakers and the listening position. The standard set up instructions also assume that the listener might be experiencing the sound at different heights and over a large-ish listening area - when nearly all of us working in small rooms are sat in chairs, with our ears kept at a fairly consistent height, and tend not to move around by more than a few feet in any direction, generally speaking. And of course, nearly all of us have a desk positioned between our monitors and our ears… and we all know about the problems of desk reflections.

Ignoring these factors when taking baseline measurements for DSP software leads to inaccuracies in the end result; and hence some adaptations are needed.

So I’m interested to know if - and especially how - any other Gearslutz have taken such issues into account when setting up DSP software in their own control/mix rooms - ?

I’m hoping to try some experiments myself, which will include:
  • Keeping the measurement mic at a consistent height.
  • Taking lateral measurements at positions along the width of my desk only (rather than also taking measurements behind my listening position).
  • Using a smaller than recommended measurement area, perhaps just a few square feet, which might better emulate my head movements when mixing (which are fairly minimal… even when working on hip hop!).

My thinking is: it’s pretty much impossible to create the luxury of a nice, large sweet spot in a typical small control room. So it might be better to focus on creating a very accurate yet much smaller sweet spot instead.

Comments very welcome

Last edited by Skol303; 5th September 2018 at 02:53 PM.. Reason: Relentless perfectionist
Old 5th September 2018
  #2
Gear Addict
 
🎧 5 years
PS: I recently experimented with using just a single point measurement in Dirac Live - i.e. taking just one reference measurement at my listening point (equidistant between the speakers) and then proceeding directly to creating a correction EQ curve.

This is the result - frequency response and decay (FR tilt is my own preference for the EQ curve). Insanely flat! But accurate only to within an area of approximately 1 square foot, requiring a surgeon's vice for holding my head in place

Room correction software (DSP) - optimising results for small rooms-frequency-response.jpg

Room correction software (DSP) - optimising results for small rooms-decay.jpg
Attached Thumbnails
Room correction software (DSP) - optimising results for small rooms-frequency-response.jpg   Room correction software (DSP) - optimising results for small rooms-decay.jpg  
Old 6th September 2018
  #3
Here for the gear
 
🎧 5 years
tagging along
Old 6th September 2018 | Show parent
  #4
Gear Guru
 
1 Review written
🎧 5 years
i've been using various dsp to correct speaker response both live and in the studio for many years - i suggest to measure in multiple locations and average results (unless your room shows very strange behaviour just a few inches outside the optimum listening area).

better rooms show less variation at varying distances (except for hf roll off further back from speakers) but one can always only get optimum results for a single position - not so much of a problem in studios but it can become a huge issue when aligning pa's for a very large crowd...
Old 6th September 2018 | Show parent
  #5
Gear Addict
 
🎧 5 years
Quote:
Originally Posted by deedeeyeah ➡️
i've been using various dsp to correct speaker response both live and in the studio for many years...
Thanks for the input

I think the issue with small rooms - at least certainly, my small room - is that the frequency response in particular can vary quite considerably over even a small area of measurement (e.g. a square metre). I find that the low end response stays mostly consistent, but that higher frequencies become impacted by comb filtering in different ways at different measurement locations. And I'm guessing that these variations will reduce the accuracy and usefulness of any averaging.

So as you say, I'm leaning towards reducing my measurement area and aiming to optimise the results around a smaller listening 'sweet spot' - much smaller than is recommended in the Dirac/Sonarworks setup instructions - but which I hope will be better suited to a small mix room such as my own. I'm going to run some tests next chance I get and will share the results here...

And yeah, I can definitely appreciate why aligning a PA for a large crowd would be whole different ball game!
Old 9th September 2018
  #6
Gear Addict
 
🎧 5 years
Ok, so I’ve now run some tests and think I’ve found what is the optimum setup for using Dirac Live in a small studio (i.e. with a small listening position). Or at least I’ve found the optimum setup for my studio. YMMV

Here’s what I discovered…

First, I set about trying to determine what affect the measurement area (i.e. the space in which mic readings are recorded) would have on the accuracy of the end result. Normally, I set up Dirac Live using the recommended minimum area of 1 square metre, but I’ve always suspected that in a small room like mine even such a relatively small area is prone to variations in frequency response that can potentially distort the end result. So I set about trying to find a compromise between:
  • A measurement area large enough to capture a useful picture of the room acoustics;
  • But not so large that it gets distorted by the ‘mess’ outside of the listening position I work in.

So I took three sets of readings: 1) a single measurement from the listening position; 2) a set of measurements over the recommended 1 x 1m area; and 3) a set of readings over a smaller 0.5 x 0.5m area. I then applied a flat EQ curve to each of these measurements in Dirac Live, hooked up Fuzzmeasure (and REW) and began testing the results.
  • The single point measurement resulted in laser-like precision at the listening position (see graph in my earlier post above), but the accuracy decreased quite rapidly once I began taking test measurements away from the sweet spot between the speakers, as you'd expect.
  • The standard 1 x 1m measurement area resulted in a coarser frequency response (i.e. more peaks and nulls), but the response varied less as I began taking test measurements away from the listening position. So it resulted in a better average over a larger area, so to speak.
  • The smaller 0.5 x 0.5m area gave the best of both worlds: much more accurate than the results of the 1 x 1m test at the listening position (almost as accurate as the single point measurement in fact), and no significant variation in response when I then tested the result away from the sweet spot - at least, no more significant than when using data from the larger 1 x 1m area.

For reference, here’s a graph showing the frequency response results of the standard 1m square area vs the smaller 0.5m square area:

Room correction software (DSP) - optimising results for small rooms-large-vs-small-area.jpg

So it seemed clear to me that the smaller 0.5 x 0.5m measurement area was the way forward in my room. Next step was to improve the accuracy of measurements…

After further experimentation, I ended up settling on mic positions shown in the crude diagram below. An initial measurement equidistant from the speakers (as always!); one either side corresponding with the location of my ears; and then a series of 6 further measurements around the 0.5 x 0.5m area. In the diagram below, the mic positions shown in blue were all at ear height; those shown in purple were of varying heights:

Room correction software (DSP) - optimising results for small rooms-mic-positions.jpg

I found this resulted in the ‘flattest’ and most consistent frequency response (once a test curve had been applied in Dirac Live). To illustrate, here’s a graph showing a bunch of test measurements taken around the 0.5m square area with a ‘flat’ curve applied in Dirac Live. Note the consistency to within +/-5dB (or less).

Room correction software (DSP) - optimising results for small rooms-listening-area.jpg

And here are the final results...

Frequency response with my preferred house curve applied:
Room correction software (DSP) - optimising results for small rooms-frequency-response.jpg

Decay times shown in waterfall and spectrogram. Note that I have a gnarly room mode at 32Hz, which Dirac Live does a reasonable job of helping to reduce:
Room correction software (DSP) - optimising results for small rooms-waterfall.jpg
Room correction software (DSP) - optimising results for small rooms-spectrogram.jpg

TL/DR: if you have a small control room then try experimenting with a smaller measurement area when setting up Dirac Live. An area of 0.5 metres square, using the mic positions shown above, worked well for me
Attached Thumbnails
Room correction software (DSP) - optimising results for small rooms-large-vs-small-area.jpg   Room correction software (DSP) - optimising results for small rooms-mic-positions.jpg   Room correction software (DSP) - optimising results for small rooms-listening-area.jpg   Room correction software (DSP) - optimising results for small rooms-frequency-response.jpg   Room correction software (DSP) - optimising results for small rooms-waterfall.jpg  

Room correction software (DSP) - optimising results for small rooms-spectrogram.jpg  
Old 9th September 2018
  #7
Gear Addict
 
🎧 5 years
Here's a copy of the REW file showing the final results, for anyone interested (see attachment).
Attached Files
File Type: mdat Skol303_DiracLive.mdat (8.37 MB, 205 views)
Old 9th September 2018 | Show parent
  #8
Gear Guru
 
1 Review written
🎧 5 years
i haven't looked at your latest files, but things seem fine from what you described earlier!
regarding the lf bump, you might try to further cut it down by changing the spacing between speakers/their distance to the sidewalls. or do you use a sub? if so, what's the position?
Old 9th September 2018
  #9
Gear Addict
 
dariva's Avatar
 
🎧 10 years
Big fan of DSP room correction here too. Since I started using Ik Multimedia's ARC, I started delivering far more consistent mixes (I think).

My not very scientific approach is to make several measurements and listen to my speaker checking playlist with each of them. I choose the one I like the most. It is a bit tricky since they will vary mostly with where they place the notch in the 400-600 Hz region and this changes the overall feel of music, especially vocal heavy tracks. But yeah....actually, it's time for new measurements, so great timing of this post!
Old 9th September 2018 | Show parent
  #10
Gear Addict
 
🎧 5 years
Quote:
Originally Posted by deedeeyeah ➡️
i haven't looked at your latest files, but things seem fine from what you described earlier!
regarding the lf bump, you might try to further cut it down by changing the spacing between speakers/their distance to the sidewalls. or do you use a sub? if so, what's the position?
Thanks!

The rise in the low frequency response is part of my house curve in Dirac Live - it’s set to +2dB at 37Hz.

If you mean the LF bump in the decay times, it’s caused by a room mode at 32Hz (and another around 50-60Hz). I’ve done pretty much all I can to reduce the low end ringing it causes: the room is heavily treated (incl. some pressure-based traps targeting 30Hz) and unfortunately due to the room dimensions, there’s no space to move the speakers any wider (the room itself is narrow and measures 5.4 x 2.2 x 2.4m L/W/H).

Yeah, I do use a sub. It’s positioned nearfield on the same horizontal plane as my monitors, placed just to one side (slightly to the left) to avoid being dead centre across the width of the room.
Old 9th September 2018 | Show parent
  #11
Gear Addict
 
🎧 5 years
Quote:
Originally Posted by dariva ➡️
Big fan of DSP room correction here too. Since I started using Ik Multimedia's ARC, I started delivering far more consistent mixes (I think).

My not very scientific approach is to make several measurements and listen to my speaker checking playlist with each of them. I choose the one I like the most. It is a bit tricky since they will vary mostly with where they place the notch in the 400-600 Hz region and this changes the overall feel of music, especially vocal heavy tracks. But yeah....actually, it's time for new measurements, so great timing of this post!
Using reference material is always a crucial test in my opinion. Acoustic measurements are of course very useful, but the ears are always the final arbiter

Haven’t used ARC myself but I am familiar with Sonarworks, which is also good - but not as good Dirac in my experience (Dirac just seems to get a better grip on the low frequency modes).
Old 9th September 2018
  #12
Gear Addict
 
YourBestFriend's Avatar
Room correction hardware-
Old 9th September 2018 | Show parent
  #13
Gear Guru
 
1 Review written
🎧 5 years
Quote:
Originally Posted by Skol303 ➡️
...due to the room dimensions, there’s no space to move to speakers any wider (the room itself is narrow and measures 5.4 x 2.2 x 2.4m L/W/H).

Yeah, I do use a sub. It’s positioned nearfield on the same horizontal plane as my monitors, positioned just to one side (to the left) to avoid being dead centre across the width of the room.
bad luck with the size/dimensions... - regarding the sub: i recommend positioning it to face it sideways so that the cone comes as close as possible to the front wall. also, do you run your mains full range and if so, down to what frequency? how about the x-over?

(reason i'm asking: i suspect the overall speaker response is affected by too many speakers going too low and being positioned at distances which lead to bad summation/partial cancellation in the lf range - let me know if i'm wrong!)
Old 9th September 2018 | Show parent
  #14
Gear Guru
 
1 Review written
🎧 5 years
Quote:
Originally Posted by YourBestFriend ➡️
Room correction hardware
very nice! how much? - i prefer this one though:

LM 44 | LM SERIES | Lake | Categories | MUSIC Group - Lake)
Old 9th September 2018 | Show parent
  #15
Gear Addict
 
🎧 5 years
Quote:
Originally Posted by deedeeyeah ➡️
bad luck with the size/dimensions... - regarding the sub: i recommend positioning it to face it sideways so that the cone comes as close as possible to the front wall. also, do you run your mains full range and if so, down to what frequency? how about the x-over?

(reason i'm asking: i suspect the overall speaker response is affected by too many speakers going too low and being positioned at distances which lead to bad summation/partial cancellation in the lf range - let me know if i'm wrong!)
Yeah, I can't do anything about the room size unfortunately... not even with the 'hardware' option shown above!

The sub is down-firing and positioned very close to the front wall, but it could be moved a little closer, like a few inches.

Mains aren't full range, they crossover with the sub at 60Hz.

Just for clarity: what LF problems are you seeing in the results? Are we talking about decay times, or the dip between 60-70Hz? (e.g. seen in the graph above, which shows a bunch of test measurements I took around the listening area).
Old 9th September 2018 | Show parent
  #16
Gear Guru
 
1 Review written
🎧 5 years
Quote:
Originally Posted by Skol303 ➡️
Yeah, I can't do anything about the room size unfortunately... not even with the 'hardware' option shown above!

The sub is down-firing and positioned very close to the front wall, but it could be moved a little closer, like a few inches.

Mains aren't full range, they crossover with the sub at 60Hz.

Just for clarity: what LF problems are you seeing in the results? Are we talking about decay times, or the dip between 60-70Hz? (e.g. seen in the graph above, which shows a bunch of test measurements I took around the listening area).
my comments are not so much aiming at the measurements but at the physical positions of your speakers:

at your x-over frequency of 60hz, you get the combined output of three speakers. you want your speaker's lf to combine in a positive way (meaning getting additional output and no destructive interference). to achieve this, the speakers must be placed at max. half of the distance of the wavelength of the x-over frequency. no amount of dsp can correct this, it's the physical position that needs to be sorted out first - response then mostly looks better from the start/without much correction.

also, if your room has some issues at about (half) the frequency of your current x-over frequency, i would at least try another frequency and/or another type/slope, all of which will create different phase behaviour.

when aligning subs and tops, phase is the single most important factor to watch, so get the tops first - and then measure the sub with the mic on the floor, not at ear height.

unfortunately, these topics often get overlooked in studio sound...

Last edited by deedeeyeah; 10th September 2018 at 12:17 AM.. Reason: wrong information given
Old 10th September 2018 | Show parent
  #17
Gear Addict
 
🎧 5 years
Quote:
Originally Posted by deedeeyeah ➡️
my comments are not so much aiming at the measurements but at the physical positions of your speakers:

at your x-over frequency of 60hz, you get the combined output of three speakers. you want your speaker's lf to combine in a positive way (meaning getting additional output and no destructive interference). to achieve this, the speakers must be placed at max. half of the distance of the wavelength of the x-over frequency. no amount of dsp can correct this, it's the physical position that needs to be sorted out first - response then mostly looks better from the start/without much correction.
Ah I get it now, thanks for the explanation

I must admit that I'd never really considered the possibility of phase cancellation between my mains and sub... doh!

So if my crossover frequency is 60Hz, that gives a wavelength of 573.33cm, half of which is 287 cm rounded up. If I understand this correctly, then my speakers (incl. the sub) should be no further than 287cm from listening position? If that's right then I should be at a 'safe distance', as they're currently positioned around 1.7m from my sweet spot. Here's a sketch of the positioning:

Room correction software (DSP) - optimising results for small rooms-speakerposition.jpg

Quote:
Originally Posted by deedeeyeah ➡️
also, if your room has some issues at about (half) the frequency of your current x-over frequency, i would at least try another frequency and/or another type/slope, all of which will create different phase behaviour.
I do also have a troublesome room mode at 32Hz, which as you note is roughly half the crossover frequency between by mains and sub. So this could be causing problems? I never knew... learning things all the time here

I can experiment with some different crossover frequencies very easily, so will try that out.

If it helps to know, the dip around 60-70Hz has been a continual problem in this room. It was there long before I installed the sub; in fact it was something like -30dB when I started out. But every little helps, so even I can save just a few Decibels by experimenting with crossover it'll be worth it.

Thanks for the input and education! Appreciated.
Attached Thumbnails
Room correction software (DSP) - optimising results for small rooms-speakerposition.jpg  
Old 10th September 2018 | Show parent
  #18
Gear Guru
 
1 Review written
🎧 5 years
Quote:
Originally Posted by Skol303 ➡️
...So if my crossover frequency is 60Hz, that gives a wavelength of 573.33cm, half of which is 287 cm rounded up.
If I understand this correctly, then my speakers (incl. the sub) should be no further than 287cm from listening position?
still not quite (i should have written this more clearly in my previous post, sorry): what happens with speakers half a wavelength (of frequency od symmetrical crossover) apart is that you get optimum coupling with positive phase relationship. a quarter distance leads to inverse phase with maximum cancellation.

so it's not so much 'up to' a specific frequency/distance but more 'precisely at a specific frequency/distance' where phase relationship and coupling can be predicted and can either lead to nice results or gets you in troubles which cannot be corrected by applying corrective eq. some phase issues (not all) can be corrected by applying frequency dependent phase inversions (allpass filters) but you need very good measurement and interpretation of data, the appropriate tools to implement and lots of experience to do this...

one way to escape potential issues though especially in small rooms without much options to move speakers is to experiment with different x-over points, slopes and also asymmetrical curves.

there's a good deal of literature available on the topic of aligning sub to tops and sub arrays within themselves also on youtube: check out anything that bob mccarthy (and a few others) published/shared.
Old 10th September 2018 | Show parent
  #19
Gear Addict
 
🎧 5 years
Quote:
Originally Posted by deedeeyeah ➡️
what happens with speakers half a wavelength (of frequency od symmetrical crossover) apart is that you get optimum coupling with positive phase relationship. a quarter distance leads to inverse phase with maximum cancellation....

...one way to escape potential issues though especially in small rooms without much options to move speakers is to experiment with different x-over points, slopes and also asymmetrical curves.
Great advice again, thanks.

From what you're saying, a crossover of 50Hz in my room would be a bad idea, for example, because 1/4 of the wavelength would be 172cm, which is almost exactly the distance from my listening position to the speakers. So I'd get max cancellation.

Whereas a crossover of 100Hz would give me a 1/2 wavelength distance of 172cm, which would result in max reinforcement... I think that's right!?

So it seems I should try pushing my mains/sub crossover higher above 60Hz (where it's currently set) and more towards 100Hz, mindful that the sub will obviously become more directional and noticeable the higher I go.

This is a great idea, especially as I don't really have any space in my room to experiment with different speaker positions. I'll try it out next chance I get
Old 10th September 2018 | Show parent
  #20
Gear Guru
 
1 Review written
🎧 5 years
all correct. give new settings a try as there is a chance that things may further improve - no guarantee though!

...or to summarize: phase coherence of spaced sources imo is much more important for optimum sound than many people tend to believe while i think that a balanced frequency response is not that important as we are pretty good in adapting to whatever circumstances we are experiencing. also, speaker response is up to personal preference although the majority of people around here believes in flat frequency response being optimal for good translation...

if i find some time, i may post a plot of the phase behaviour of a typical lr 4th order 3-way crossover - you might be surprised!
Old 10th September 2018 | Show parent
  #21
Gear Addict
 
🎧 5 years
Quote:
Originally Posted by deedeeyeah ➡️
all correct. give new settings a try as there is a chance that things may further improve - no guarantee though!
I’m on it... Will report back here once I’ve tried some alternative crossover settings.
Old 12th September 2018 | Show parent
  #22
Gear Maniac
 
🎧 10 years
Quote:
Originally Posted by Skol303 ➡️
Here's a copy of the REW file showing the final results, for anyone interested (see attachment).
There is a curious bump in distortion between 160 and 300 Hz in the left channel (also seen in left+right). A little odd that the left and right IR peaks have opposite polarity.

As an aside, there was a small fix to the V5.19 build a few days after the initial release, your file looks to have been made using the original build so worth re-installing to get the latest.
Attached Thumbnails
Room correction software (DSP) - optimising results for small rooms-left.jpg   Room correction software (DSP) - optimising results for small rooms-right.jpg  
Old 13th September 2018 | Show parent
  #23
Gear Addict
 
🎧 5 years
Quote:
Originally Posted by JohnPM ➡️
There is a curious bump in distortion between 160 and 300 Hz in the left channel (also seen in left+right). A little odd that the left and right IR peaks have opposite polarity.
Thanks for the input John! From the ‘horse’s mouth’ of REW, as it were

Yes, I’ve noticed those discrepancies between left and right distortion and IR readings myself, and it’s left me scratching my head for an answer to what might be causing it. Any ideas from what you can see in the measurements?

I otherwise have the following ‘hunches’…

Distortion: The subwoofer is positioned closer to the left speaker than the right; and there is something nearby in the room that resonates/vibrates at a low frequency during the measurement sweep (possibly my desk; or it could be the radiator that is fitted to the wall immediately behind the sub and speakers). The resonance is more audible/pronounced when I measure the left side in comparison to the right. Could this be the culprit of the curious distortion reading?

Impulse Response: this one has me stumped... there's a noticeable peak in the right-hand measurement that is absent from the left, but I've never been able to pinpoint the source of it. My room is set up mostly symmetrical (plan view included below for reference - measures 5.4 x 2.2 x 2.3m L/W/H). If not the room, could this potentially be a fault with the right-hand speaker itself?

Thanks again for any input.

PS: updated .mdat file attached with more recent measurements (DSP EQ set flat for ease of comparison).

Room correction software (DSP) - optimising results for small rooms-plan-view.jpg
Attached Thumbnails
Room correction software (DSP) - optimising results for small rooms-plan-view.jpg  
Attached Files
File Type: mdat 180912_Skol303.mdat (8.36 MB, 82 views)
Old 13th September 2018 | Show parent
  #24
Lives for gear
 
🎧 5 years
Quote:
Originally Posted by Skol303 ➡️
T If not the room, could this potentially be a fault with the right-hand speaker itself?
I had this case with an xlr cable.
Old 13th September 2018 | Show parent
  #25
Lives for gear
 
akebrake's Avatar
 
🎧 10 years
Quote:
Originally Posted by JohnPM ➡️
There is a curious bump in distortion between 160 and 300 Hz in the left channel (also seen in left+right). A little odd that the left and right IR peaks have opposite polarity.

As an aside, there was a small fix to the V5.19 build a few days after the initial release, your file looks to have been made using the original build so worth re-installing to get the latest.
Interesting! Thanks John PM for chiming in

When comparing expanded impulses I saw a difference between L FLAT and R FLAT. Are there any special "signature" too look for?
Usually I don't look for Phase jumps at 11 kHz...

Best regard

Akebrake


@SKOL

What speakers are you using now? Any DSProcessing in this frequency area?

One way to show what John points out (I think) is plotting the phase curves as unwrapped Phase (Lin freq. scale) ovly
Also overlaid are older SKOL measurement (without phase jump).
Attached Thumbnails
Room correction software (DSP) - optimising results for small rooms-phase-comp-flat-l-vs-older-meas.jpg  
Old 13th September 2018 | Show parent
  #26
Gear Guru
 
1 Review written
🎧 5 years
duh...

these graphs as shown in post #22 and #25 look horrendous! - i'm not checking mdat-files (cause i'm much more familiar with other measuring tools) so i didn't notice that there must be some rather serious issues which rend any further attempts of getting a reasonable overall system response pretty much useless unless they are getting fixed!

here's how: put the speakers on the floor, the measurement mic too, bring the capsule down as close as possible to the floor (to avoid any reflections). measure the speakers (each on it's own) with the mic 1m away. leave the mic where it is, change speaker (and use laser, marker etc. for precise positioning), use sweeps, pink noise and bursts, don't change the slightest bit regarding, cable, volume or mic gain - then compare: they should measure within a very narrow range, not more than half a db off. (if not, throw them away!)

then put them back into their original position, put the mic at ear height in the listening position and measure individually, again pink, sweeps and bursts - they still should measure pretty much the same, maybe a little more off (say within 1.5 db). if they don't, use just one output of the daw (could it be that one d/a converter went bad?) and the same cable.

hope this will get you back on track...

by the way: when aligning the sub to the mains, put your mic on the floor! at ear height, it cannot 'know' whether it 'hears' direct or reflected sound: bad for measuring/ phase aligning. also, do not use sweeps for aligning the sub, only pink noise (and not filtered/band limited for reasons that would lead too far to be discussed in this thread).

but do the mains first, including all eq corrections - oh, and i'm still not sure about the position of your sub: not on the floor, firing downwards? not a good starting point! think about wavelength/phase/distance and what happens when the sound bounces back from the floor...


p.s. do dirac live and/or sonarwork allow for setting multiple allpass filters?

Last edited by deedeeyeah; 13th September 2018 at 09:09 PM.. Reason: edited and p.s. added
Old 13th September 2018
  #27
Gear Addict
 
🎧 5 years
Quote:
Originally Posted by dinococcus ➡️
I had this case with an xlr cable.
Interesting. So the problem could potentially be caused by an irregularity in the XLR cable? I would never has suspected… and will investigate. There’s certainly no noticeable (i.e. audible) difference between the two speakers, but I’ve learnt not to trust my ears alone.

Quote:
Originally Posted by akebrake ➡️

What speakers are you using now? Any DSProcessing in this frequency area?

One way to show what John points out (I think) is plotting the phase curves as unwrapped Phase (Lin freq. scale)
Akebrake! Great to have your REW wisdom brought to bear on this problem

In answer to your questions:
  1. My speakers are Unity Audio Rock MkII mains and Avalanche Sub.
  2. Yes, I’m using Dirac Live DSP, implemented using a miniDSP DDRC-22A hardware box in the signal chain immediately before the speakers (hooked up 2.1: L/R into the sub and then out to the mains, crossing over at 60Hz in my first set of measurements and 100Hz in the second set). Note that Dirac Live imposes an ‘inter-speaker correction delay’ on the output, which is intended to compensate for differences between each speaker and the listening / measurement position. The software version allows this delay to be switched off, but the hardware version - which I’m currently using - doesn’t.

Apologies for my ignorance, but can I confirm which set of measurements you think is ’best’? The first set (labeled LEFT / RIGHT) or the more recent set (labeled FLAT L / FLAT R)?

I attempted to replicate your own REW ninja skills and recreate the graph that you attached above… which for me was an education in itself

I assume that viewing ‘unwrapped phase’ is no different to interpreting any phase relationship - i.e. the closer the measurements for each speaker, the better? If so, then my second set of measurements here (showing FLAT L & FLAT R)…

Room correction software (DSP) - optimising results for small rooms-measurements-2.jpg

…appear to be an improvement on my initial measurements here (LEFT & RIGHT):

Room correction software (DSP) - optimising results for small rooms-measurements-1.jpg

Or I have I got that wrong!?

Quote:
Originally Posted by deedeeyeah ➡️
these graphs as shown in post #22 and #25 look horrendous!
Good thing I like a challenge!

In all seriousness, some very good points in your post and good advice which I'll try out.

Re. "do dirac live and/or sonarwork allow for setting multiple allpass filters?"... in all honesty I'm not sure, mostly because I'm not so familiar with all-pass filters (other than phaser plug-ins!) as I'm very much an amateur and still learning. I think the difference between Dirac and Sonarworks is that Dirac uses mixed phase correction rather than minimum-phase and linear-phase room correction. But I could be wrong.
Attached Thumbnails
Room correction software (DSP) - optimising results for small rooms-measurements-2.jpg   Room correction software (DSP) - optimising results for small rooms-measurements-1.jpg  
Old 13th September 2018
  #28
Gear Maniac
 
🎧 10 years
Left and right impulse responses have the same polarity peaks in the "Flat" measurements, though both are inverted.

Resonance doesn't produce distortion rises like those in your measurements. Probably worth measuring with Dirac bypassed to see if the same behaviour is apparent. If not, Dirac may need a little more headroom allowance. If it is still there might be a drive unit problem, but worthy of more investigation in any case.
Old 13th September 2018 | Show parent
  #29
Gear Guru
 
1 Review written
🎧 5 years
Quote:
Originally Posted by Skol303 ➡️
...my second set of measurements here (showing FLAT L & FLAT R)…


there still must be a different filter setting in the hf range on the dirac between channels to get this result: it's almost impossible that this difference is caused by the room! cannot imagine the two speakers to be off that much either.
flat meaning through the dirac but with flat settings? did you compare to physically bypassing dirac, so: cr out to speakers? (and quickly checking the cables for polarity should be easy...)
Old 14th September 2018 | Show parent
  #30
Lives for gear
 
🎧 5 years
Quote:
Originally Posted by Skol303 ➡️
Interesting. So the problem could potentially be caused by an irregularity in the XLR cable? I would never has suspected… and will investigate. There’s certainly no noticeable (i.e. audible) difference between the two speakers, but I’ve learnt not to trust my ears alone..
I do not confound polarity and phase. For me this is two things differrent. JohnPm had used the word of polarity.

Dirac plays with the phase and this is a thing had displaise to friends who test the trial version.
📝 Reply

Similar Threads

Thread / Thread Starter Replies / Views Last Post
replies: 289 views: 36015
Avatar for Kraft
Kraft 4 weeks ago
replies: 96 views: 9325
Avatar for Mikael B
Mikael B 30th September 2017
replies: 1005 views: 125299
Avatar for MogwaiBoy
MogwaiBoy 28th April 2021
Post Reply

Welcome to the Gearspace Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…

Forum Jump
Forum Jump