mEQ Equalization System

mEQ Equalization System
"The powerful yet cost effective mastering equalization system...

mEQ is a high performance low cost mastering equalization system that runs under the Windows family of operating systems. mEQ introduces a new approach to the problem of mastering equalization by combining linear phase digital filtering with spectrum analysis to give unprecedented levels of control over the sound quality of a mastered recording.

So why is mEQ better than conventional approaches?

To answer this question we first need to define conventional approaches. Current practice for mastering of popular music will typically involve a combination of processes including mix-down, dynamic range compression, equalization and normalisation. mEQ addresses the issue of equalization. After the mix-down the final mix will more often than not have some level of dynamic range compression applied to it, whether that is through a conventional compressor or a multi-band compressor like that available in the finalizer. After compression some degree of equalization may be needed to compensate for the shift in spectral balance caused by compression, or worse still, to compensate for a poor mix-down. Either way, the way in which this is achieved is by an operator (of varying skill) adjusting the controls on a graphic, or more commonly a parametric, equalizer while monitoring through speakers to obtain the desired sound. The whole process is heavily reliant upon the degree to which the operator can judge the music and its deficiencies : in other words how well he/she can analyze the recording through listening. Even for the most skilled of operators this task can be particularly difficult to do well because of the way we perceive sound. Factors such as masking, poor monitoring acoustics, or poor speakers play a significant part but most importantly the issue of human perception of loudness makes the task very difficult.

Human hearing is volume dependent so at low volumes we find it hard to hear low and high frequencies. So if you simply turn up the volume you may find that to you the mix sounds better. If an operator, for example, raises the mid-range by 6dB as he thinks it sounds better how much of this improvement is due to better spectral balance as opposed to a mere increase in overall volume (because by raising the mid-range by 6dB you will have raised the overall volume of the recording by something a bit less than 6dB). To accurately judge a process like this requires that the operator reduce the overall volume whilst boosting the mid-range, but the question is by how much should the volume be attenuated. As can be seen the process becomes very difficult to handle using the conventional approach.

So how does mEQ handle this problem? mEQ is essentially composed of two parts : a spectrum analysis engine and a high resolution linear phase digital filter. mEQ handles the problem by first analyzing the entire recording given a measure of the average and peak spectrum content (displayed graphically) from which the operator can quickly and easily judge the spectral balance of a given recording. Then through a novel user interface it allows the user to design a matching digital filter by adjusting the average spectrum to the form that the operator would like to achieve. When designed the operator can listen to the effect of the filter and accurately compare it with the un-equalized version in real time by switching the filter in and out during playback. Now how is this different from the conventional approach?

Firstly, you do not need golden ears to judge where the problem areas of a particular recording are as the average and peak spectrum measurements give you an accurate indication of where those problems are. Should you find this hard to believe try this for a challenge. Find what you believe to be is your best sounding CD and your worst sounding CD and analyze each in turn. You will find the differences quite obvious. With more and more experience (from seeing good and bad recordings) you will come to a better understanding of the relationship between a particular average spectrum shape and the overall sound quality.

Secondly, a typical equalization filter like those found in the finalizer will have only a limited degree of flexibility (ie. limited boost & cut, Q etc.) and are unlikely to be linear phase filters. The filter in mEQ is an 8192 point linear phase FIR filter whose characteristics are designed to match and compensate for the average spectrum as closely as possible. The degree to which you can boost and cut a particular region is essentially unlimited.

Finally, following on from the discussion of loudness, mEQ attempts to compensate for the effect of changing apparent loudness due to filtering by measuring a loudness figure for the orignal recording (based on an A-weighted sound power) and adjusting the volume control on the filtered version so that the loudness figure is the same as the original. In this way the operator cna be sure that the differences he/she is hearing is due to the change in spectral balance and not the change in volume.

A typical Equalization example

As an example of the flexibility and use of mEQ consider the following example. We shall analyze the equalization of two different tracks on two different CD compilations both by the same artist. In one case the recording is very listenable whereas in the other rather fatiguing (the authors opinion). Figure 1 below shows the spectrum display of mEQ for the track My Wave by Soundgarden taken from the album Superunknown. The top trace is the peak spectrum for the track whereas the bottom trace is the average spectrum and the middle trace is the geometric mean between the two. In general the spectrum is well balanced with perhaps a little too muhc energy around the 5kHz region (quite common for the grunge / metal genre).

mEQ screen shot

Figure 1 - My Wave by Soundgarden

Now contrast this with the spectrum shown in Figure 2 which is for the track Mind Riot by the same artist taken from the album Badmotorfinger. In this case although the region around 5kHz seems to be better control there is a large hole in the region from 400Hz to almost 2kHz which makes the recording sound thin and without body.

Given such a recording then how can we attempt to correct the imbalance in the spectral content? From figure 1 we can infer the general shape of an acceptable recording for this genre of music. General features include about 5dB emphasis in the region from 150-300Hz, a generally flat spectrum up to 8KHz and then a gentle roll off with an attenuation of around 10dB at 20kHz. Also the spectrum should be as smooth as possible but care should be taken so as to not introduce strong peaks in the peak spectrum as these may sound disconcerting (vocal resonances are are common issue). Using this knowledge and the gain cursor of mEQ we adjust the desired spectrum to that of figure 3.

Figure 2 - Mind Riot by Soundgarden

In this case we have adjusted the spectrum to conform to the desired shape but constrained by the desire not to introduce any strong resonances in the peak spectrum. As a consequence it was necessary to leave a few discontinuities around 250Hz and 1.5kHz. Note that the filter is designed by dragging the control points using the gain cursor. The control points are spaced at 1/3 octave intervals and are constrained to move vertically (in analogy with an ideal 1/3 octave graphic equalizer). Note also, that the control points can be shifted to peak power, average power or median spectrum plots and in this case we have moved it to the peak power plot to aid in adjustment. If after listening to the result we are satisfied with the filter we can equalize the source and record it to a new wave file using the record button.

Figure 3 - Mind Riot by Soundgarden, Equalization design

Now you may be wondering how accurate our filter really is and whether the new response is indicative of how the spectrum will change after filtering. One easy way to check is to record the result and open it with mEQ which will produce a new analysis for the equalized recording. Figure 4 shows just such a test. Note the close matching between figures 3 and 4. The only significant departure occurs at around 200Hz with the actual equalized result showning a minor peak whereas our filter design does not show one. This is most likely a resolution problem of the filter (that is it is only an 8192 point FIR). Note that mEQ is only in alpha and this issue will be investigated further in the comming months.

Figure 4 - Mind Riot by Soundgarden, Equalized recording

Conclusions

By coupling spectrum analysis with linear phase digital filtering it is possible to provide a powerful, yet funtionally simple means of re-adjusting the spectral balance of sound recordings. This scheme has potential for both the mastering of new original compositions as well as re-mastering existing recordings which show problems. How mEQ is best used is something ill defined at this point in time owing to the infancy of the idea. With time and experience a better understanding of this issue will no doubt come about.

Downloads

The current Alpha version of mEQ has expired. At this point in time a free alpha is not available. It is anticipated that a new pre-release version with restrictions will be made available. Much of the information in this article is very out of date and need for revision. Due to time constraints the earliest an update can be expected is four weeks time. Apologies for any inconvenience.

"We use Zeus for Windows as our development editor of choice..."

Paavo Jumppanen
Author of mEQ Equalization System