25 Measure Twice, Listen... Forever? Studio Acoustics Measurements
The Perfect Studio
The purpose of this (and many) studio builds is to create a great listening space, a place where one can mix and/or master audio with complete clarity and translation to the 'outside world'. To achieve this, we firstly need to know what constitutes a great listening space, an answer that has evolved throughout the years of studio design, but is often vaguely, and perhaps over-simplistically, defined as a room with a flat frequency response. Meaning all frequencies playback (at the listening position) at the same sound pressure level (dB SPL)... but is this what we really want?
The Flat Frequency Deniers
More often than not, we don't actually want to listen to a completely 'flat' system, and there have been several studies suggesting that we actually prefer, or at the very least more accustomed to, listening to music with a bass boost and a tilt to roll off the high frequencies, this is where room/speaker curves such as B&K curve originate. In 1974 Bruel and Kjaer conducted a study that measured the response of Hi-Fis in typical home listening environments, ie. how the general public consumes music, rather than how speakers sound in lab conditions. By analysing the frequency response of 5 sets of loudspeakers in 3 rooms, recording responses from human listening tests, then combining the data with pre-existing concert hall response curves they devised the optimum curve shown below. You can read the study itself here.
From this point on, many recording studios were designed to have their playback systems replicate the B&K curve. Since then, there have been further studies measuring listeners' playback preferences such as Harman's 'Listener Preferences for In-Room Loudspeaker and Headphone Target Responses' available at AES. Olive et al. (2013) found that 'The preferred in-room loudspeaker target response is not flat but has a bass boost of about 6.6 dB below 105 Hz and a treble cut of -2.4 dB above 2.5 kHz.' The black line on the graph below shows the preferred listening curve of speakers (headphones in cyan) from this study.
With the combination of consumers' preference and control over tonality, as well as speaker and headphone manufactures designing products with these curves in mind, designing studios with a perfectly flat frequency response appears to be fruitless. Creating a space with possibly not the most pleasing listening environment, and mixes unable to translate as convincingly to the real world, is not what we want! But often, when someone says a flat-frequency response, what they most likely mean is a smooth frequency response, with as few peaks and nulls in the room's response as possible.
Let's Face Facts
In reality, achieving even this kind of flat or smooth frequency is nigh on impossible, as there are too many influencing factors that can't be removed from a normal studio environment; the desk or console, instruments, furniture... the floor even. These elements will all interfere with soundwaves and contribute to the 'sound' of your studio, such as introducing comb-filtering at certain frequencies. But, as long as we establish an aim, we can try to get as close as possible, and as they say, shoot for the moon...
Only Half The Story
However, be aware that the frequency response isn't the whole picture, reverberation plays an important role in how a room sounds, and importantly, reverberation across the frequency spectrum. Too large a reverb time would create a cloudy, unclear environment, making mixing decisions harder, while too dry a space will sound unnatural and could lead to confused mixing decisions. It is arguably better to have an even spread of decay times across the frequency spectrum than it is to have a flat SPL response. Another aspect that can sometimes be ignored is the accuracy of the stereo field; we want to hear the same response from both the left and right speaker (or more if thinking about 5.1 or Atmos rooms). Being able to control all these factors is part of the art of studio design.
The Why and How of Measuring Acoustics
This is where the importance of taking acoustic measurements comes in; we can take initial readings, see where the problem areas are, track the progress of the build, and see if changes are working. Although we can trust our ears to a certain extent, we can't be expected to recall how a room sounded before some treatment was put in with any accuracy!
Taking acoustic measurements is a fairly simple process using a free piece of software called Room EQ Wizard or REW for short, an amazing program that can take readings of your room and display the results in a whole host of ways. As long as you've got an omnidirectional microphone, and an audio interface, you're good to go. Fortunately, I had a measurement microphone from my previous Sonarworks purchase, but there are some cheaper alternatives available such as the Behringer ECM8000.
REW will play a sine wave sweep out of your speakers, and the microphone will record the response. I won't explain exactly how to set it up here, as there are already lots of great tutorials online, such as this one from GIK Acoustics. It is best to record each speaker one at a time, and always from the same microphone position (unless experimenting with different listening positions). For best results, you can also measure your audio interface and upload the calibration file of your measurement microphone to ensure any anomalies in your equipment do not influence the recorded results.
Acoustic Readings and Reading Acoustics
Now that we've taken a measurement, what are we actually looking for? Well, these are three main graphs that we'll look at in this blog: Phase, SPL, and the Waterfall graph.
SPL - Displays the frequency response of the room, with decibels on the y-axis and frequency on the x-axis. This is the graph we tend to want to see a 'flat' or smoother line on. (Fig. 1. Seen in dark red on the top)
Phase - Displays the phase position of the sine wave, ideally, we want to see a smooth line transitioning through 0-360 degrees over and over again as the sine wave cycles. This display is 'wrapped' so the line will jump back to the top once a cycle is complete. (Fig.1. Seen in light red on the bottom)
Waterfall - Perhaps the most useful, and the one I'll refer to most in this blog, shows the SPL response, but crucially, also related to time. So, for how long each frequency 'rings out' in the room after the initial playback has occurred. The x-axis shows frequency, the y-axis shows SPL, and the z-axis shows time in milliseconds (fig.2). Ideally, we want a smooth initial SPL recording followed by a gentle, even decay across the whole frequency spectrum.
REW also displays some other useful graphs such as distortion, RT60, and spectrogram, but I won't get into them in this post.
A Note on Scale
Before we get into the actual measurements, I just want to highlight some issues when reading frequency response graphs, and a feature microphone and speaker manufactures often use to their advantage! This is the scaling, and smoothing of data. Below are three graphs, that are all the same reading but with different scales and smoothing. The first is smoothed to 1/48 dB per octave and zoomed in on the y-axis, the second has been smoothed to 1/3 dB per octave at the same zoom level, and the third has then been smoothed and zoomed out. Now, which one would you use to sell the idea that your room/microphone/speaker has a flat frequency response?
For most of the measurements I will show, they are for the right speaker only, apart from the final results. This is just to save on doubling up pictures and keeping things simple. Smoothing will be set to 1/48 dB per octave when showing SPL and phase, but using psychoacoustic smoothing for all other instances - as this algorithm correlates more closely to the perception of human hearing.
Reference Measurements
Let's begin with some base measurements, so we know what we're potentially aiming for. Below are some measurements from an example studio, which was professionally designed and built by an acoustic design company.
We can see that there is still a reasonable amount of peaks and troughs across the frequency spectrum, but the decay time across the board is fairly consistent, lasting around 130ms as we can see in the waterfall graph below. The phase is relatively smooth throughout with a couple of anomalies at 220Hz, 300Hz, and 1.7kHz, which you can see translates to nulls in the frequency spectrum.
The main problem areas are the large disparity in the room response below 50 Hz, with large peaks at 25 and 45Hz, a null at 38Hz, and a long decay time at 25Hz. A rather large peak at 150Hz is also a concern and something to be aware of when working in this room.
Now let's compare the example studio to a typical home spare room/bedroom studio with no acoustic treatment...
As we can see, although the frequency response doesn't look too bad in and of itself, the phase and waterfall graphs reveal the many acoustic problems of this room. The phase is frankly all over the place and there are large inconsistencies between how long different frequencies reverberate, which is going to massively cloud one's judgment when mixing.
And this is only the right speaker! If we combine this with the results of the left speaker, we'd see that the response from each speaker varies a lot, meaning the stereo image is going to be a mess as well! In the example professional studio, each speaker has a similar response, resulting in a clearer stereo image.
Basement Studio Results - Before
Now let's take a look at what I was working with! Above is a reminder of the empty basement room, and below are the initial readings. I think you'll agree, these might be some of the worst room acoustic measurements you'll ever see!
So, it was definitely going to be an uphill struggle, but I still needed a goal in mind of what I wanted to achieve:
A smooth frequency response
A tilted response, somewhere between the B&K and Harman Curves, essentially a 6-8dB tilt
Even decay times across all frequencies, ideally not too short
A smooth phase response
Ok, let's see how I got on!
Acoustic Measurement 1 - Bass Trapping
It was at this stage, once the speakers were mounted in the walls, that I could take the first set of measurements. Most of the bass-trapping had been done at this point, along with the ceiling, functionally, if not visually, and floor. The front wall wasn't finished at this point and was pretty much an empty shell, the diffuser was yet to be installed, and some broadband absorption was missing. With no speaker calibration, here were the results...
As you can see, a vast improvement from the initial measurement, with the reverb time coming right down to around 150ms for most of the frequency spectrum, and a much more even response overall, however, the response still has some large peaks and nulls, and the frequency response is too flat for my liking.
Acoustic Measurement 2 - Initial Speaker Calibration
At this point, I experimented with the Neumann KH310's onboard speaker controls to get a more even response. We can see a much more unified low-end and the peaks aren't quite as drastic as before.
Acoustic Measurement 3 - Finished Front Wall
A few changes for this measurement. First of all, the paneling on the front wall was finished - meaning a denser front wall acting more like an infinite baffle. A rug was moved in, although not sure how much difference this made to the readings! Then further speaker calibration to make up for the extended bass frequencies from the soffit mount. Speaker EQ:
-2.5dB Bass
-4.5dB Low Mid
0db Treble
Again, slight improvements, the mid to high end is getting smoother, and there is a lot less ringing in the low-mid range.
Acoustic Measurement 4 - Diffuser
For this measurement, a diffuser and more absorption were added to the back wall.
Some further smoothing out of mid-range, and the peaks and nulls are just getting less extreme. Small improvements here and there!
Acoustic Measurement 5 - Further Tweaking
Fluffy bass absorption was added behind the front wall, completing the transition to a semi-solid/absorptive soffit wall. Further tweaks to the speaker calibration:
0dB Bass
-4.5dB Low Mid
-1dB Treble
The room response is getting there now, more even across the board and with a gentle tilt.
Final Raw Acoustic Measurements
