Without A Q

We live in a world where microprocessors regularly present complex information to us in simple terms. In fact, the complexity of many daily tasks is completely forgotten, as our view is focused on the result and not on the software assisting in the process. However, in many current audio processors the complexity remains visible while the results may be difficult to see. It may be understood that technical people will be dealing with the equipment: people willing to suffer the arcane terminology and clumsy visual metaphor. This may add a certain esoteric cachet to the technician's tasks, but rarely does it lead to better sound, installed at less cost.

Audio professionals needn't suffer the vague and misleading terminology common to consumer audio equipment. They don't need to convert values in their head, either. The most common example of obscuring the task with the user interface is the use of a Q-factor to represent bandwidth in a parametric equalizer. This is not a problem limited to software interfaces; there are analogue hardware equalizers that make the same mistake. In either case, the intention of the user is to set the range of frequencies affected by the filter being adjusted. The information needed while adjusting the bandwidth of a parametric filter is the range of frequencies that will be affected. Few people have an intuitive understanding of the exact range a specific value of Q will provide.

The Q-factor of a filter is defined (very concisely by Bob Metzler in the Audio Measurement Handbook) as the ratio of the centre frequency to the bandwidth measured at the -3 dB point. The result of a Q setting of 10 for a filter set to 1000 Hz is a bandwidth of 100 Hz. Seems simple enough, until the centre frequency is 5325 Hz and the Q is set to 7.25, and then it may be more difficult to calculate the bandwidth (734.48275862 Hz) in your head. Perhaps the more direct bandwidth-value, in octave fractions would be easier to understand. After all, we have spent much of our professional careers dealing with octave fractions, enough to make them seem almost intuitive. Of course, this is the same assumption the engineers make that design equipment with Q values instead of bandwidth values. They have used Q to define the nature of a capacitor in a filter circuit and the Q-factor of the resulting filter. It is very natural for them to use Q as the definitive value for controlling the width of a filter. The design engineer, naturally, tries to express the internal workings of the device on the front panel controls. Unfortunately, they are not the best judges of the user interface, because they are looking at it from the wrong side of the box.

A good user interface will provide the necessary controls to make appropriate adjustments in the field during the final setup of the system. This includes providing the appropriate legends on each control to support this task. In the case of a permanently installed sound system, the end user is the technician doing the final system adjustments. This person views the signal processor as a black box, which offers the necessary processing and range of adjustment. There is no interest (at this point in the project, anyway) in the inner workings of the box. The technician's perspective is looking from the inside of the system (a collection of black boxes) out to the acoustical response achieved in the venue. The controls and their legends should be defined from this point of view.

During the system setup, the technician measures the acoustical performance to allow the precise adjustment of the parametric equalizer. The on-screen display of the measured frequency response will be in Hz. On-screen cursors may offer the ability to measure the width of each significant frequency response deviation in Hz. This is less difficult to calculate based on the concept of bandwidth in octave fractions (400 Hz = 1/3-octave at 1200 Hz) than the concept of Q-factor. However, both require the user to perform calculations.

When using more sophisticated acoustical measurement systems (MLSSA, Smaart, TEF, etc), it would be far more convenient to set the filter skirts (lowest and highest frequencies that the filter will affect), as this is the information most readily available from a measurement system. The result would be a parametric equalizer that is set by adjusting the filter bandwidth directly, instead of indirectly. We shift from adjusting the parametric filter centre frequency and Q/bandwidth to setting the low and high frequency range of the filter. This would finally relate the setting of the EQ to the information available from the measurement system. We would have an interface that directly relates to the perspective of the user, rather than the designer of the equipment.

United Entertainment Media Inc.
460 Park Avenue South, 9th Floor
New York, NY, 10016
Ph. 212-378-0400
FAX 212-378-2160

page construction