Hard proof that equalization kills room modes
In recent years there has been much debate about how effective equalization is in reducing the negative impact of room modes. Room modes are characterized by a peak in the frequency response and extended ringing in the time domain. They create low frequency coloration and lack of articulation and are perhaps one of the most prevalent acoustical problems, since they exist in all small rooms. This article provides measured evidence of the efficacy of equalization in reducing both the peak and ringing associated with room modes.
Equalization should be thought of as the last component of the optimization process. First an appropriate acoustic design should be completed taking into account the use of the space (two channel listening room or home theater; number of seats; family room vs dedicated, etc), the radiation characteristics of the speakers and the number of subwoofers permitted. Once the design is complete installation of the acoustic treatment and equipment needs to take place. Next is the process of calibration, where the acoustical targets for the level of reflections, modal decay, frequency response and RT60 are confirmed by measurements and speaker placement, levels, delays, phase and crossover slopes adjusted. Finally comes equalization, when all other adjustments have been made.
The correct application of equalization to address room modes is not a straightforward subject. Historically equalization filters were calculated based on a simple inversion of the frequency response, with no consideration as to the location / Q of modal resonances, the correctability of the issue or the location of speaker boundary interference related suckouts. It is probably these simple correction filters that have caused the debate around the efficacy of equalization.
The theory behind correction of room mode resonances is described clearly by Floyd Toole:
“Room resonances at low frequencies behave as “minimum phase” phenomena, and so, if the amplitude vs. frequency characteristic is corrected, so also will the phase vs. frequency characteristic. If both amplitude and phase responses are fixed, then it must be true that the transient response must be fixed – i.e. the ringing, or overhang, must be eliminated” (Toole, The Acoustical Design Of Home Theaters, 1999)
There has been debate about whether this theory applies in small rooms and also whether small rooms have any areas that are minimum phase. On the latter point, John Mulcahy, creator of the acoustic measurement package Room EQ Wizard has written a useful article that explains how a measurement called excess group delay can be used to understand which areas of a room’s response are minimum phase. On the former point we present the measurements in this article as proof that the ringing can indeed by reduced or eliminated by correctly deployed equalization.
The following measurements were taken in a dedicated 9 seat home theater designed by my friend and acoustics accomplice Jeff Hedback of HdAcoustics. On this project we split the workload – I did the calibration and Jeff did the acoustic design. Practical limits of space and cost limited the amount of low frequency absorption that could be deployed, so it was known from the outset that equalization would be appropriate in helping to improve performance. The design was completed in such a way that the seat to seat variation in frequency response was low, allowing equalization to be applied effectively. It was confirmed during the verification measurements that a couple of parametric EQ filters would be beneficial in improving the sound quality in the room. Two equalization filters were applied, at 25Hz and 50Hz. Since room modes have a very narrow bandwidth – typically around 5Hz – you need narrow bandwidth or high Q filters to properly address them. The filters used had a bandwidth of 0.35 octaves. The attenuation at 25Hz was 13dB and at 50Hz was 6dB.
Before the equalization was applied there was a slight subjective loss of articulation of piano, bass guitar and drums due to the resonance at 50Hz. Furthermore there was also a significant ‘energizing’ of the room at 25Hz such that low frequency effects from movies were slightly blown of out proportion and were even making the screen noticeably shake! After the equalization there was a very clear improvement in articulation and that energizing of the room at 25Hz was almost completely nullified, creating a very articulate and dynamic bass presentation. Certainly one of the better home theaters I have had the pleasure of calibrating, and a great testament to the combined skills of a home theater installer, a specialist calibrator and a specialist acoustic designer all playing their part in achieving the results.
Three different measurements are presented: the frequency response, the time/energy/frequency response and the impulse response. The frequency response measurement clearly shows how the equalization filters have flattened out the response. The time/energy/frequency measurement, presented both as a waterfall and a spectrogram, clearly shows the reduction in decay times. The impulse response measurement most clearly shows the reduction in ringing.
Time / energy / frequency
Proof indeed that properly applied equalization can reduce both the frequency response peaks and time domain ringing associated with room modes.
In response to Ethan’s comment about the response at other locations…
The home theater had 9 seats. One of the important things about acoustic design for home theaters is to minimize the seat-to-seat variability in bass response through acoustic treatment, subwoofer placement and seat placement. If seat-to-seat variability is low then equalization can be effectively applied. Equalization, remember, is global in effect, in that it applies to all seats in the same way. During the acoustic design Jeff Hedback, of HdAcoustics, modeled a number of subwoofer locations with the aim of minimizing seat-to-seat variability, including opposing walls and floor placement. In the end two subs were placed behind the screen about mid-way between floor and ceiling.
The following graphs show frequency response and before / after spectrogram plots at position B, which is towards the front right of the room. The previous position measured was inbetween seats C & D (the two seater couch was changed to a three seater). The relative levels of the two graphs does not fully reflect the change due to the two parametric EQ filters, since other adjustments were made to levels during the calibration process. For reference both measurements were taken over 20Hz to 20kHz and levels set to 75dB SPL using a C weighted meter. In any event it is the shape of the curve and the +/- error that is important and looking at this you can see that equalization did not make things worse and the spectrogram charts are much improved.
Update on 2015-09-21 18:14 by Nyal at Acoustic Frontiers
I can’t believe we are still debating this….
Let me present the data in a different format…I am posting normalized spectrograms. These are great for showing the rate of decay as they are normalized to the peak energy at each frequency.
Only two filters were applied, one at 25Hz, the other at 50Hz.