Myths and facts about studio acoustics III: Auto-calibration

Italian acoustician Donato Masci from Studio Sound Service concludes his series of articles on studio acoustics with a look at the co-habitation of auto-correction with acoustic treatment.

3.1 Auto-calibration

There are many hardware and software systems on the market that allow you to calibrate studio monitors — some of them are installed into the monitor itself. The most advanced auto-calibration systems record — by placing a calibrated microphone at the listening position, or at different positions — an impulse response of the room through a test signal (typically a sine-sweep) for each individual monitor. From the impulse response it is possible to obtain a lot of information including those in the time domain and the fine phase alignment between the individual speakers. In particular, the system also calculates the frequency response at the listening position — or the measurement position or positions — and then uses some EQ filters to try to improve it.

3.2 Frequency response problems

 As said in my previous articles, the frequency response is essential to analyse the quality of a control room but it must be considered together with other useful parameters, and, of course, with critical listening. In the frequency response (FR) you can see most of the problems but these are often overlapped with others.

Problems in the FR may be different in nature but can be divided into two main groups. The first depends on the acoustic field of the room (involving the reverberation and the wave-acoustics — room modes) while the second depends on all the phenomena related to the early reflections (the most energetic ones) with boundaries or other surfaces in the room. In general, when speaking about the first type of phenomena the reverberant field above 200Hz colours the FR in a very smooth way, emphasising those frequencies for which it has more reflection from the walls. This effect is practically the same at every point in the room. On the other hand, the room modes — the effects of which can be seen mainly below 200Hz — behave much more aggressively on some individual frequencies and the influence on the FR is related to the position of the source and the listener in the room.
Even the phenomena related to reflections (or rather the interaction between direct and reflected waves, in general) are more peaked for some frequencies and depend on the position of the source and the listening point because they are based on the distances between the direct and reflected paths (as explained in my second article).

After writing the first of this series of articles in which I reported the statistics of some of the acoustic parameters measured in 60 recording studios where I worked, I met Aki Mäkivirta and Christophe Anet of Genelec and read their article A Survey study of in-Situ Stereo And Multichannel Monitoring Conditions presented at the 111th AES Convention in 2001, which shows the measurement data of 372 studio monitors in 164 top control rooms around the world. All the rooms analysed here had big monitors and on average were notably bigger than those I considered in my measurements. Apart from the data on reverberation and other acoustic parameters the most striking section of their article is about the FR. Considering the relative notch at frequencies below 1000Hz in the third octave smoothed frequency response, the result is that ‘the median notch depth is 14.2dB, but 30dB notches are not uncommon’ and that ‘the most typical notch frequency is 100Hz, but deeper notches appear at higher frequencies.’
Knowing how to read the FR is one of the most daring jobs for an acoustician because it is not intuitive to imagine how many dBs will change in the FR after a slight variation in the room (such as moving a monitor) but at the same time it is sometimes unnerving to note what little change there is when we introduce an acoustic trap. This is due precisely to the fact that, as we said before, the phenomena that contribute to the ‘colour’ of the frequency response are different in nature.

3.3. Case study: auto-calibration of the same monitors in different rooms (acoustically treated and untreated)

To better explain my experience with auto-calibration I’m referring to some project studios I’ve recently worked on. The idea is to compare two control rooms with the same type of monitor with auto-calibration (Genelec 8260a) — the first was acoustically treated (even if it is a very small room and full of acoustic compromises) the second was completely untreated.

Figure 1

The reverberation times are extremely different in the two rooms, despite the fact that they are roughly of the same size. In particular, the untreated room has a reverberation time @ 63Hz that is five times longer than the treated one. The Centre Times in the untreated room are too long at low frequencies, while the chart for the treated room is perfectly placed on the average values that I showed in the first article of the series (V13.2). As you can see from the chart, both parameters are not substantially changed by the calibration.

Figure 2

From the frequency response of the untreated room, a fundamental problem immediately arises — a particularly narrow notch of about 10dB @ 53Hz due to the interaction with the corner and the rear wall. In addition, frequencies close to the notch are particularly emphasised (30Hz and 80Hz) by the room modes. The rest of the FR has several comb filters –- problems in the mid and mid-high frequencies that a trained eye can easily see — however it remains within +/-3- 4dB from 200Hz to 20kHz, confirming the fact that an untreated room often has a FR that could seem good if it is not carefully analysed (actually, midhigh frequency comb filters like these are very annoying, because they are those to which the ear has maximum sensitivity).
Auto-calibration is able to control some of the reflections at medium frequencies (even if it fails to improve the strong reflections at 800Hz and the comb filter between 2-8kHz) and to balance the sound energy on the low and mid-low frequencies, making listening more controlled — but paradoxically the notch at 53Hz is slightly worse.
The listening is improved and everything seems certainly more balanced (for example, you do not notice the big-drum booming effect anymore), but for the low frequencies you can’t really understand what is happening nor the precise frequency. Remember that the reverberation time was obviously still extremely long.

Figure 3

In the treated room the result is different. Although this is a small room, with room modes particularly close in the spectrum, it is equipped with some acoustic treatment to control low frequencies (especially on the ceiling, on the back and on the sides) that make listening conditions good. Auto-calibration has, in this case, brought the room to a professional level and substantially improved the linearity of the mid frequencies (there were reflections from the mixing desk and the outboard) and controlled all the resonances and the ‘minimum phase low frequency boost’.
In a nutshell, the collaboration between auto-calibration and acoustic treatment has allowed the customer to have a professional full-range and perfectly balanced listening experience in a room with a height of 2.5m and an area of less than 20sqm.

Figure 4

3.4. What can auto-calibration improve?

What I have noticed in my experience, is that many auto-calibration systems work very well for mid-high frequencies, focusing the monitors and giving the right brightness even in listening environments where the room does not have the right diffusion.
It also works well for all problems related to the interaction with hard surfaces, such as large consoles and racks, which are typical of small control rooms full of outboard close to the listening position. These problems are generally between 500Hz and 2,000Hz.
On the other side, at low frequencies the effectiveness is closely related to the problem. Good auto-calibration certainly works very well in containing the low frequencies that are emphasised by the proximity of a hard surface, such as when monitors are close to the wall or are in-wall — in other words when you have to make equalisation to correct a minimum phase low frequency boost. If, however, there are strong non-minimum phase effects, such as those created by boundary reflections, auto-calibration has problems. In these situations, different software behaves in different ways, and, in my opinion, the ‘smarter’ software ‘understands’ when it is better not to intervene.
Another phenomenon that generally creates problems for auto-calibration is the modal resonance of a room, because it is closely related to the measuring point. In this case, the SPL at the frequency of the room mode changes from a maximum to a minimum (with easy variations of 20dB) maybe in a few centimetres, and clearly, if the system is based on a single measurement point, it could optimise the listening in a very small area, making it worst in the rest of the room. For these reasons I generally prefer to use a multipoint system.

3.5. Dispelled myths about acoustic treatment vs. auto-calibration

Let’s conclude by returning to the myths I set out in my first article (V13.2) and address those two final subjects.

8. Auto-calibration is useless if a room has good acoustic treatment — False. What I have noticed is that auto-calibration systems give the best of themselves in situations, such as home and project studios, where you have physical limitations on being able to guarantee results using only acoustic treatment. They also work very well for fine-tuning ‘minimum phase low frequency boost’ and to control reflections on the mixing desk and the outboard, which is very useful even in more professional situations with in-wall speakers. 
I have to mention the subwoofer. After everything I’ve said in previous articles, if the phase and the level are not properly calibrated, it is probably better not to have a subwoofer. The worse the room acoustic is, the more difficult it is to have proper subwoofer integration. Our ear is so imprecise at LF (remember that at 35Hz listening to a sinusoidal signal, you need 9dB of level change at the audio source to subjectively hear and perceive a level change…) that proper subwoofer integration in a room is not straightforward without measuring equipment. In reality, most subwoofers are totally uncontrolled. In this case auto-calibration (and in particular auto-phase) is really useful.

9. I don’t need acoustic treatment if I have auto-calibration — False. Today auto-calibration systems are not able to solve all the acoustic problems of a room, because, by their very nature, they are not able to manage some of the phenomena that create the problems themselves. As I hope to have shown in the case study, auto-calibration does not replace proper acoustic treatment.


Donato would like to thank Christophe Anet and Valentina Cardinali, for useful discussions and thoughts, and his friend Christopher Martinuzzi for reviewing the text.