DRC, with open baffle speakers

[otto88]

I’m planning an XO that's PC based, so I’ve been reading through this http://www.diyaudio.com/forums/multi-way/63078-how-pc-xo.html
It makes sense to allow for flexibility & future uses. at this stage choosing the sound card (almost picked the Saffire PRO 40)

In particular, ShinOBIWAN and ewildgoose’s experiences have convinced me that after the XO, to minimise room problems I’ll want to add/ try DRC.

A good summary of the benefits of DRC from that thread are here:
http://www.diyaudio.com/forums/multi-way/63078-how-pc-xo-29.html#post823319

I hope/ expect that with very good drivers (and source and amplification) and a *lot of time spent fine tuning, the speakers will be really good.

However the speakers will be open baffle, which have less room interactions.

How much benefit would/ does DRC provide with open baffle speakers?

And what “extra” do you need include in the sound card spec, so it handles DRC? DRC say on left, right, and centre: 3 extra inputs and outputs?

 

[otto88]

mic?

(Reading on) ShinOBIWAN found after using a Behringer EMC8000 mic "fairly poor"

"I personally wouldn't use DRC without a top notch mic unfortunately"

I'm also in process on a similar $ (cheap) mic, the Dayton EMM-6, but properly calibrated from
Cross·Spectrum - Calibrated Dayton Audio EMM-6 Microphones for Sale

The Premium-Plus is measured individually for:
• on-axis response
• frequency response at 45° and 90°
• polar response (for measuring room acoustics)
• and sensitivity and noise floor

Is a EMM-6 adequate for DRC, if properly calibrated?

 

[catapult]

Is a EMM-6 adequate for DRC, if properly calibrated?

Yes IF your measurement software can use the calibration file supplied by CSL. Hopefully your DRC software will be able to import the corrected impulse response or frequency response from software like HOLM or ARTA that can use the cal file.

 

[otto88]

Thanks catapult, I suspected that: calibration is the key.

I haven’t yet chosen the DRC software, but importing the corrected response should be a prerequisite.

 

[Reverberant]

Is a EMM-6 adequate for DRC, if properly calibrated?

This is Herb from Cross-Spectrum Labs, a poster asked me to weigh in.

As catapult noted, you first have to be sure that whatever DRC package you use can accept mic calibration files. There are a couple that do (REW comes to mind).

As for the measurement itself, most mic cal files are for using the mic in an on-axis configuration, where the signal is assumed to be coming straight on along the long axis of microphone (mic pointed toward the sound source). In room acoustics applications, there will be direct sound from the speakers, but there will also be sound waves from reflections arriving at random directions to the microphone. Since 1/2-inch microphones are not omni-directional at the highest frequencies (above ~ a few kHz) using the on-axis correction file could cause some problems at the higher frequencies.

That's not to say that you can't get good data using the on-axis cal files, it's just that you may (and I stress may) have to work a little harder at it. I started offering 45/90-degree calibrations for room acoustics applications to avoid the problems some of these problems. But you can certainly experiment a little with on-axis cal files and get perfectly acceptable results.

One thing about DRC measurements is that you may find that even with the mic kept in the same spot, that you get different results every time you run a measurement. One thing you can do to try to avoid this is spacial average - you can take measurements with the mic in different locations in the listening area and then generate an average of the resulting curves, or (my preferred method) put the mic at the end of a long stick and slowly move the mic in horizontal and vertical circles around the listening area during the sweep.

HTH

 

[gainphile]

I've heard DRC few times and including DEQX2496 and all leave me cold. The resulting eq is simply unnatural to the ears. I think this is because you can't eq time domain with amplitude correction.

The best speakers I've built (OBs) were the ones measured flat in freespace (outdoor)

But I don't mind correcting one or two peaks in the bass region for room modes.

 

[63781]

I agree with gainphile here. In severe cases DRC will improve things, that is when the speakers are very far from flat or the room is really bad.

My approach is to correct the *speaker* with EQ above 2-400 Hz or so, and tailor the speaker to the room with EQ below 2-400 Hz. That could be named "manual DRC"

 

[otto88]

Yes, bass and esp low bass is where most of the problems are.

I think the DEQX2496 is probably DRC built to a price; good value, but with some compromises. Maybe there’s something about it’s processing, which creates that sense of being unnatural.

The time domain? Good point, I don’t know yet if software eg Bidule also works in the time domain??

What DRC did you try on the OBs, the Behringer?

"manual DRC" may give 80% of the benefit, for 20% of the complexity

 

[63781]

Bidule cant do much on its own - it needs DSP plugins. If you use an impulse response convolver you can do things in the time domain also.

Behringer DEQ2496 is very limited in its EQ capabilities - 10 parametric EQ's is simply not enough. Its automated EQ is .. well... no good at all! Better to set it manually assisted by proper measurements I think.

Proper EQ software can do much more - the number of EQ points is only limited by computer CPU power. Sofware EQ can also be Linear Phase, which none of the hardware EQ boxes can do.

 

[otto88]

I didn’t at first see Reverberant’s post

(gainphile, was the mic used properly calibrated for DRC?)

REW has quite a following, with its own forum REW Forum at Home Theater Forum and Systems - HomeTheaterShack.com - could well be worth trying.

I can see why a sweep of the listening area would be good.

Though why with the mic in the same spot, might you get different results every time you run a measurement; do they vary much?

 

[otto88]

Stig

“the number of EQ points is only limited by computer CPU power”

Thanks, glad to learn that!! For the XO-EQ-DCR I was going to get just a modest PC: better get one with some grunt.

REW (Room EQ Wizard) Room EQ Wizard - REW Home Page does eg “spectral decay plots, waterfalls and energy-time curves; generating real time analyzer (RTA) plots” - Is this properly addressing the time domain?

 

[63781]

Also note that different EQ software plugins use different amounts of CPU. In general - LinearPhase EQ is very CPU heavy, while EQ's that are not LinearPhase are much lighter.

The worst example of CPU load is probably PLparEQ, which I use myself. Its also the most smooth sounding EQ I've heard...

 

[63781]

REW (Room EQ Wizard) Room EQ Wizard - REW Home Page does eg “spectral decay plots, waterfalls and energy-time curves; generating real time analyzer (RTA) plots” - Is this properly addressing the time domain?

No, it does nothing in the time domain. DRC cannot do that either, since the room decay is very long - several hundred milliseconds. You might be able to clean up the first 10 ms with impulse response convolution, but you cant do anything with the actual room decay.

 

[gainphile]

Interesting read here:

Frontiers

Since DRC can only correct a very small window in space, what about the other infinite number of spaces. They have direct effect to sound due to reflection etc. The DRC will not able to correct deviations on all space based on finite number of sweeps (or worse: single location in sweet spot).

Quote :
"The issue with room equalization is not about sufficient computing horse power,
but how and where to apply it. "

The "where", I believe, is the lower frequency region to surpress room modes. I don't want to be sounding like a naysayer so take this with a skepticism too and go ahead and share your results! That is what the "Y" is for in DIY

 

[David_Web]

The best room correction is done in the brain.
I would make the speakers behave as nicely as possible without a room and then let the brain do the rest.
Maybe some broad EQing to get a balanced response.

At LF it might be a bit different but at 80+ I would leave room correction to the brain.

To record the IR I would use HOLMImpulse and to process it I would use DRC (the command-line app) Has plenty of options to correct for what the user wants.

 

[Jaco Pastorius]

Using Sbragion’s DRC you can correct part of linear distortions introduced by speakers and room interaction. You could create a minimum phase filter or a mixed phase one so you can also correct part of what is called “time domain”.

Obviously the system is “space dependent” so it can be done only for a single sweet spot.

 

[wowo101]

Has anyone used Dennis Sbragion's DRC for driver and crossover correction instead of full room correction yet? As far as I understood the software's inner workings, it should be quite usable for that if you take nearfield measurements as input, correcting only the speaker's "own" linear and phase distortions.

Any experiences?

 

[Jaco Pastorius]

The purpose of Sbragion’s DRC is to correct part of linear distortions introduced both from the system and from the room. When you listen to the speaker in the sweet spot you have linear distortions caused by the system and the room so why correct only the first one?

It’s also impossible to measure only speaker transfer function without an anechoic chamber.

 

[otto88]

I should see what sort of PC power, DRC needs

Thanks for the reality reduction on expectations

 

[wowo101]

Jaco - the idea would be to use DRC as an automated means of EQ and phase correction for the drivers (instead of doing it manually), and leaving out room correction because of its limitations concerning the listening positions.

Concerning measurement limitations: Sure, but I'm working within the limitations of indoor nearfield and outdoor farfield measurements when taking measurements for manual EQ and driver alignment, too, so I guess there's not a lot more error to be expected than I already have in my current process...