Updated: Mar 2, 2020
If you've dealt with room acoustics, and geeked out about the subject as much as I have, you'll still be frustrated by every single little frequency peak or null (room modes) in your measured frequency response of your room. It's no small task to tune a room for perfect acoustic response. It's even more futile to attempt using a software or hardware EQ solution to correct acoustic issues without first properly addressing the acoustics of the space beforehand.
There are a few good acoustic correction software platforms that you can spend a little or a lot for and only resolve very minor issues, if not create more of them. Sonarworks is a solution that has been quite publicly known and tried. Some have found good results. Many have not. Why? Let's just say room acoustics don't fully fall into the perfect science that you'd expect in every case. Plus, a perfect room isn't always perfect.
A perfectly flat and completely dead room of any acoustic reverberation or reflection of any kind sounds awful to the human ears. In fact, it's rather spooky-sounding. I won't go into room acoustics in general because there are plenty of resources available by way of Google search and Kindle books. I'm going to focus squarely on Sonarworks Reference 4 and how to get the most out of it.
I've had great success with Reference 4 Systemwide and the plugin. My system is an audiophile loud-speaker set up, which is excellent for a mastering studio, but many of you have speakers and subwoofers more tailored to your work. Unfortunately, using separate speakers for lows and mid-high ranges raises the complexity a bit.
To be clear, these are helpful tips, not exactly a full-blown perfect solution for everyone. I just hope I can help users potentially arrive at a better result.
1. Address your room acoustics and speaker placement first
Only use software/hardware correction/calibration solutions for the last 10% or less of your room acoustics issues. I repeat: use it last! You will find yourself totally dissatisfied with the results of your room acoustic profile if you don't.
Seek out GIK Acoustics who can offer some free advice. Use REW (Room EQ Wizard) to see how bad the acoustics are first. Note that, it's ok if your higher frequencies roll off a bit from 1Khz at 0db to 20Khz at -6db. But below 1Khz, down to 25Hz, the room should be quite flat. I'm sorry to tell you that I spent a few thousand on my acoustics. It was well worth the expense, however. Since I took advice and knowledge I collected, it proved to be effective.
Why? Because, if you don't address basic acoustic issues and you use software to correct the signal in a bad room, you could destroy your speakers and your ears trying to force them to be flat! Yep. you can actually over-compensate room issues with your speakers and damage them (smell those speaker coils heating up?). You can also fatigue your ears in ways you don't expect.
Another reason is, the more correction Sonarworks has to use, the quieter your Reference 4 Gain will need to be without nasty distortion. You'll have to crank your system's volume that much more.
Make sure your speaker placement is optimal. You'd be surprised at how much of a difference a few inches more from your walls might make, or a few more inches from your ears. Spend a good weekend working with them and measuring in REW. It's so worth the extra effort. Don't be impatient. Be diligent!
2. If you use a subwoofer: Get two. And then, align them as close as you can first
Also not good news, but many users of room correction software are dissatisfied because the software is trying to correct too much by way of phase and frequency cancellation, having not being properly informed where the sub is located. Sonarworks Measure certainly doesn't have a a function to detect and measure where your sub is.
If you look at a set of audiophile speakers that slightly lean backward, there's a reason for this. It's not simply to point the speaker drivers right at your face, although that doesn't hurt. The purpose of that type of speaker is so that each speaker (tweeter, mid-range, and woofer) produce frequencies that arrive at your ears at exactly the right time. Note the center of each speaker and how it's perfectly aligned vertically to the centers of the other speaker drivers. It's subtle, but produces a clearer image. Now imagine that your speakers are 10 feet from your ears but your subwoofer is only 5 feet from your ears. That's not remotely accurate. If your desk is near a wall, your speakers are near-fields directly in front of your face, and your sub is under the desk on the floor---you're already at a disadvantage.
More, to have a single sub in the middle of the room isn't, by itself, a problem but in the crossover frequencies between your speakers and your sub is where Sonarworks is going to need to correct for room acoustics issues more than most other frequencies. If you only have one sub how can Sonarworks possibly correct any phase issues around those crossover frequencies? You really need two subs and they should be placed as close to your speakers as possible.
The subs need to be aligned with the other speakers as close as possible. There are some creative ways to do this, but the one I prefer is to use two impact signals in a DAW that are precisely aligned in the session. An example would be to record a thumpy/poppy metronome directly onto a DAW track through a wired loopback in your audio interface or by using a short cable from output to input. Use that newly recorded metronome to play back through the speakers (disable the internal metronome and play back only the recorded metronome).
Set your crossover on your subs to the highest supported frequency (you can return the crossover frequency to the previous setting afterward). Now, turn off the subs. Play and record the metronome through the main speakers and record with a mic onto a new track. The microphone should be where your face would normally be at your desk. Next, turn those off and power on the subs. Record the metronome through the sub onto another separate track.
Completing this you should see a result of the timing between the originally recorded metronome (from the loopback) and the recorded metronome between the speakers and subs. If the sub and speaker tracks aren't aligned then you have some work to do. Adjust the location of the subs or speakers so that both signals arrive at the mic at the same time when the speakers and subs are all powered on. Get these as closely aligned as you can, then proceed.
3. Ditch the cheap measurement mic
You get what you pay for, and this is no exception. I'm aware you're not recording music through this mic. It's just being used for measuring frequency sweeps. True, but have you seen your microphone's frequency response, and if so, when you apply the calibration to Sonarworks Measurement it will reveal the EQ response curve that will be applied to the mic's frequency response to make it flat.
Did you catch something there? What are we using room correction software for? Then why are we having to use correction on the microphone itself, which in turn is used to measure/correct the room?
If you're serious about accuracy/measurement, the microphone you use should be serious as well. I used an Earthworks M23 measurement mic in my tests. You can rent one with a calibration file from companies like micrentals.com.
According to Earthworks, it's quite difficult to calibrate a microphone properly below about 700Hz. Do do so judiciously you'd need an an-echoic chamber the size of a football field! The reality is, calibration in the bass frequencies, where you're likely going to hear the most trouble in your room, is even more difficult with an inaccurate mic. And the calibration file is a loosely sampled compensation file that's rounded off by the measuring software. If you look into the calibration file, there's perhaps 100 sampled frequencies. Sonarworks may use far more than that for your room correction.
Fortunately, Earthworks mics (and I sound like a sales pitch) have an extremely flat response down to about 3Hz! In my case with the M23, it's also measured flat up to 30Khz, even though it's only spec'd to be flat to 23Khz. Because of this the calibration can be a mere guess between 10Hz and 700Hz. With a flat-response mic you will still be within a really acceptable margin of error.
Again, if you're serious about room calibration you should take the microphone seriously as well. In my case of using the M23, I can say that there was a noticeable difference, especially in the top end. The result I heard from a sub $100 measurement mic with calibration profile vs. the M23 was brighter--so bright that I feared burning up my Dynaudio tweeters. The M23 result was much more natural.
4. Put your computer display on the floor, facing up and out of the way
If your speakers aren't fully visible (at least 60 degrees spread) behind your screen, do this. Some points you measure in the software might hide the speakers behind your screen, and this will produce an inaccurate result.
For some, this is a little challenging if you have short video cables and such. The problem is displays cause so many acoustic issues in the higher frequencies. During measurements you want the microphone to have actual visibility of the speaker drivers. Even if you return the display back to its previous location and audio is no longer accurate, it's just a result you'll have to adjust to. After all, your ears are going to hear the result, not a microphone. You don't want your speakers' tweeters blasting super loudly to get sound through the back of your display!
I set my display in a chair further away from me, setting my display resolution lower to still see Sonarworks Measurement software instructions. Besides, if you do this, you'll not have to remeasure your room if you decide to move the display to the side of your desk.
5. Measure your listening position to speakers and your speakers to each other
The software will attempt to detect (and it's quite good at this) the distance between your speakers and to your listening position. However, measure this yourself as well. If it's way off, assess what might be causing the clicks and such not to be measured properly. Try a different test tone.
This goes without saying in Step 1, but if you have one speaker 5" further away to your listening position to the other, then your listening error of margin is distorted.
6. Be accurate
The software gives you the impression measurements should be close. Hand-holding the measurement mic is pretty vague for scientists. I suggest, for example, when you're measuring the speaker up close at about 1/2" be exactly the same on both speakers within a millimeter. When you're going through the 37 points of measurement, try and get the mic as close to the center of the new test location as possible each time. If you're struggling to do so, pull the mic away to the side and come into the circle again. The software will work with you here. It's rather patient. Even if it gives up on you it allows you to pick right up where you left off.
7. Review your result
Don't be upset if you feel the need to repeat the whole process. If the measured result is rather off-looking between the two speaker curves, or showing a severe compensation of frequencies to more than 6db, you may need to go back to Step 1. Measure again to be sure but address that problem if it's repeated.
I did the test 3 times, myself, and chose the result that felt the most natural to my ears. Then I ditched the other results.
8. Extend the Limiter
If you're like me and use speakers that can play frequencies below 25Hz and above 17Khz, go into the Limit Controls and set the high and low limits (frequency cutoff points) to "Extended". If you have big speakers that can provide super high/low frequencies, the default settings will sound small. Also, review your resulting correction curve and select "-6db" boost if only the extreme highs or lows are significantly boosting as a result. This setting can protect your woofers and tweeters from too much boost. Note the cap in my high-frequencies below.
9. Use the B&K 1974 Target Curve and Mixed Latency
These settings are optional. I use them both because they allow the speakers to sound more natural, and closer to what Bob Katz uses in his results of his software. I feel, however, the B&K 1974 curve is more natural to my ears. For Latency I use the Mixed mode because it's close to Linear-Phase, but perhaps will reduce some pre-ringing effects. Use Linear-phase only if your result curve is fairly flat. If not, avoid it.
NOTE In case you're curious, the very top photo illustrates the results of my speakers in my room using Sonarworks. Earthworks M23 microphone photo is copyright of Earthworks Inc., and used for illustration purposes only.