R/E/P Community

Please login or register.

Login with username, password and session length
Advanced search  

Pages: [1]   Go Down

Author Topic: How flat is flat?  (Read 1676 times)

Jack Schitt

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 648
How flat is flat?
« on: October 07, 2005, 12:30:37 pm »

I was test driving monitors last night. Alesis M1 Mk2's, KRK V6 series 2 and Mackie HR624's. The KRK's seemed more 'articulate' for lack of a better word but were at the same time very hard on the ear's. Almost painfully bright in a sense. These monitors easily let the snare drum smack you silly. There was detailed to be heard for sure but I'm really thinking ear fatigue would be a real issue with these speakers. I know these are popular with many but I don't think you could mix for long before your perception began to change from the pummeling the ears are taking. Owners, I would love to hear your take on it.

The HR624's on the other hand were very very smooth sounding. By comparison the snare drum for example seemed significantly reduced in level but were very smooth and kind to the ears. I've read some say the mackies make everything sound good. I can see that being a problem but is it really that bad? Again owners, your view?

The Alesis M1's weren't as loud and perhaps just a little less defined in the low end but were far closer in overall tone to the Mackies than the KRK's. I've mixed a lot of stuff on original MK1 version of this speaker so maybe I'm just more familiar with this model. Owners who have maybe switched care to chime in?

It seems pretty obvious that these speakers can't all truly be 'flat' but how far off do you have to be before that term no longer applies? I would do very different mixes on those KRK's than I would the other 2. If we can debate monitors this different in sound its no wonder people argue analog vs digital to the death.  
Logged

josh

  • Full Member
  • ***
  • Offline Offline
  • Posts: 191
Re: How flat is flat?
« Reply #1 on: October 07, 2005, 04:59:16 pm »

It's pretty easy to make a speaker that'll meaure roughly flat, within a couple of dB anyway, on-axis with just one measurement mic.

That is, if frequency response is all that's important.  There's phase response, distortion, impulse response, damping, and off-axis response in there too.

There are plenty of PA speakers that look ruler-flat on the plots and don't exactly sound that way.

Anyway, YMMV, but there's a lot more than frequency response coloring the sound of monitors.

hargerst

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1458
Re: ...
« Reply #2 on: October 16, 2005, 04:51:41 pm »

A flat speaker (to me, anyway) is one whose on-axis response curve will fit inside a +/- 2dB window.

Now, having said that, it also means there can be as much a 4dB variation (at some frequencies) between two speaker systems that meet that spec.  They won't sound the same.

The big problem that makes most of these specs meaningless is the contribution of the room.  When you get close enough to eliminate the room, most nearfield speakers are wanting in the bass range and phase problems at the crossover points become more obvious.  Move back a bit to smooth those problems out, and room nodes rear their ugly head.  

What's the solution?  Try to understand how your particular speakers/room color the sound and learn how to adjust from there.  The only other solution is the one mastering guys use: spend a ton of money on flattening the room, and then buy some Dunlavy speakers, at $40,000 a pop.
Logged
Harvey "Is that the right note?" Gerst
Indian Trail Recording Studio

Jack Schitt

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 648
Re: ...
« Reply #3 on: October 16, 2005, 06:14:30 pm »

hargerst wrote on Sun, 16 October 2005 16:51

A flat speaker (to me, anyway) is one whose on-axis response curve will fit inside a +/- 2dB window.

Now, having said that, it also means there can be as much a 4dB variation (at some frequencies) between two speaker systems that meet that spec.  They won't sound the same.

The big problem that makes most of these specs meaningless is the contribution of the room.  When you get close enough to eliminate the room, most nearfield speakers are wanting in the bass range and phase problems at the crossover points become more obvious.  Move back a bit to smooth those problems out, and room nodes rear their ugly head.  

What's the solution?  Try to understand how your particular speakers/room color the sound and learn how to adjust from there.  The only other solution is the one mastering guys use: spend a ton of money on flattening the room, and then buy some Dunlavy speakers, at $40,000 a pop.



Fair enough. Thanks. I've done what I can with the room and its several levels of magnitude better than before treating it. I have no illusions its any where near flat though. I've been using my current monitors for about 5 years now and I do know them well. Probably best to stay with monitors I know instead of making a lateral move to something different but not better.
Logged

brandondrury

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 703
Re: How flat is flat?
« Reply #4 on: October 17, 2005, 04:30:49 am »

Who gives a damn if the monitors are flat?  I think accuracy is a crock too.  I want a monitor that forces me into a corner and forces me to fix all the problems in my mixes.  If I need monitors that that color the sound in a way that makes my snare drummer better (on other systems) then I would expect that color to be a good thing.  

Since we all hear different things, I think it's impossible to use words like "flat".  It comes with some sort of assumption that we are all chasing the same frequency response, transient response, etc from our monitors.  That is simply untrue.

The big boys fight about monitors on here all the time.  That's because a given monitor gives one dude exactly what he needs to do a kick ass mix and it doesn't give the next guy what he needs to do a kick ass mix.

Brandon

hargerst

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1458
Re: How flat is flat?
« Reply #5 on: October 17, 2005, 01:03:11 pm »

Yes, and no, Brandon.  

If you're making decisions about the tone or level of an instrument, based on a specific speaker bias, that's a bad thing.  As I've said, over a period of time, you can learn to compensate for those biases, but it's easier on a flat system that's giving you an honest representation of the sound.

Kinda like driving a car that keeps pulling to the left, and you hafta double clutch it to change gears.  It'll get ya there, but it doesn't make the driving experience fun or easy.
Logged
Harvey "Is that the right note?" Gerst
Indian Trail Recording Studio

josh

  • Full Member
  • ***
  • Offline Offline
  • Posts: 191
Re: How flat is flat?
« Reply #6 on: October 18, 2005, 05:43:13 pm »

Well here's the deal:

we've all been hanging around in rooms that are about 500 square feet with a mixture of reflective and absorbtive surfaces and listening to music on nonlinear speakers for most of our lives.  This is a familiar environment and we are going to approach any speaker with biases we are used to until we untrain them.

If you play a speaker with flat axial response in a room like this then it's going to sound bass-heavy and boomy because the bass is not as directional as the high end, so you get a phase-incoherent mix of direct and reflected bass coming to your ears, and proportionately less reflected high end and fwiw the reflected high end is not a phase issue but a haas-effect masked thing psychoacoustically so you don't notice it.  So flat-response speakers will sound boomy.

So we go about putting up bass traps and making the room sound a lot less like a regular 500sqft room but it makes flat speakers sound more flat and that's all good.  Eventually we can build a very good room where flat-sounding speakers sound flat.

At that point, you have to remember that your ear is connected to your brain and a heck of a lot of what you hear gets filtered by your brain anyway.  So the psychoacoustic effect, which is not measured by a measurement microphone, is important to understand.  Also, remember that a frequency response plot is usually done with a sweep of frequencies without regard for phase, time-domain, or distortion that may be present if there's more than one frequency being played on the speaker, not to mention dynamic compression.

Well your ear/brain can quickly adapt to frequency response irregularities as long as they are reasonably consistent.  For example, if I snuck up and adjusted the EQ on your iTunes software for -3dB/octave above 3kHz then obviously you'd have a very not-flat response coming out of your headphones.  But while you might notice this for a minute or so, quickly you'd adapt and everything would sound fine, until you took the cans off and suddenly everything sounds overly crisp for a minute or two until you re-adapt to natural sound.  Barring any really outrageous frequency response anomalies, such as a complete notch filtering in a frequency band (which I have actually seen as crossover artifact), or a big 6dB peak at the port resonance, or whatever, you can pretty much adapt to any normal speaker's frequency response curve as long as you adapt to it using familiar material.

Your brain uses phase and time information for imaging choices, and crossovers, which in large part affect frequency response of a speaker, affect the phase in huge fashion, and you can't adapt to that.  So the phase/timing information that gets blurred or distorted by the crossover can't be compensated for, and the imaging will be hosed.

Also non-axial (off-axis) response of a speaker is important if you have reflective surfaces.  And speakers distort...  So you may play 1kHz into the thing and get only 1kHz out of it at 80dB but if you have a whole spectrum of varying material coming out of it at 92dB then you probably are going to have intermodulation, phase, and harmonic distortion all at once screwing things up.  This doesn't show up in a frequency response plot but it does show up in your mixes!  

So my point is that speaker design is an art and science that requires balance.  So when you evaluate a speaker response curve it all has to be in balance against the effects of fixing the respnse curve.  A midrange bump between 2K and 3K is a big deal because it's the top of vocal range detail but it's also critical for imaging, so if to fix it would require inducing a phase error in that region, so that voices cannot be properly localized in the stereo image, would be a huge mistake.  It's preferable to have the frequency anomaly than the phase anomaly.  However building a speaker with dead-linear phase response but a notch at 2kHz that's about 100 Hz wide and 30dB deep is too much emphasis on phase response... you gotta give up phase response in order to get back the frequency response...  and they do have that effect.  I have swapped capacitors in crossovers before and made no measurable effect on frequency response but dramatically changed the imaging of the speaker and overall character.  So what I'm getting at is frequency response is really the wrong thing to look at, and all to easy for a manufacturer to manipulate and convince you the product is better than it really is.

You have to listen to the speaker and figure out if it's any good with your ears, and listen for all of these details.  Forget about frequency response...  all modern studio monitors are close enough, and even some have DIP switches (like our JBL LSR's) to change it.  If the manufacturer of our $2200 monitors thought frequency response was such a holy grail would they really make it so easy for a recording engineer to change it on a whim?  Even if they don't have switches, you change it by placement in the room or if you're like me, you crack it open and hack on the XO to make it do what you want anyway.  At some point it may even not be bias, but taste.  I don't like an overly bright speaker, but I like a more "forward" midrange and I will screw up mixes if I don't get speakers that do this...  but my Alesis M1A's sound much more "forward" than our LSR28's but the pink-noise plot is nearly identical...  Subsequently I have switched the 2kHz switch to "+2dB" on the LSR28's and suddenly my other engineer is pumping out mixes with way too much sub-bass...  hey one man's treasure I guess...  It's not the frequency response so much as the way our brains try and compensate for it.  Too much 2K makes one guy want to compensate with 50 Hz, and too little 2K makes me want to pan everything to the middle.  It's not as predictable as you might think!  It'd be fine if your brain were not involved...  but heck all of us have imperfect ears and brains that have been used all of our lives to compensate.
Pages: [1]   Go Up