Speaker DSP and room correction is a hotly debated topic. When not done properly it can certainly do more harm than good but I believe that when done correctly it will always be better than not having it in the system. In order to achieve the highest quality systems, DSP should be used in a properly treated room. Having said that, fully implementing a DSPed system is a significant expense compared to a traditional system.
In my system I use DEQX units to handle this. There are a couple of other options on the market but none which are as extensive or with a fully integrated setup process. Some alternatives include the Acourate programs, some of the miniDSP devices, and Trinnov which provides more limited capabilities. Despite which system you use, implementing it correctly is critical and should be done by an expert. Recently I’ve been handling system calibrations on my own but only after having seen the process done by Larry Owens and Ken Goerres a number of times.
The DSP in my system targets 4 different areas. The first is the room correction in the lower end of the spectrum. Most people will be familiar with this type of correction. Because of minimum phase behavior below the Schroeder transition area, a simple EQ can correct this region. Of course an analog EQ would cause various degradations to the system so ideally it should be done with DSP before any stage of conversion. One issue which people normally run into is trying to overcorrect their low end which results in more harm than good. This form of room correction must be used in combination with acoustic treatment. Peaks can be brought down by a considerable amount but dips should not be brought up by any large amount using DSP due to the fact that any additional energy at those frequencies will cancel itself out. Even rooms with large amounts of acoustic treatment will have variations in the low end. The main purpose of the acoustic treatment is to bring up the dips as well as to control the decay time. Once this has been achieved, the DSP handles the last bit of flattening the response curve.
The next area which the DSP in my system treats is the higher frequency correction of my speakers. This is largely a correction of the driver response on its own but factors some room anomalies in as well. Many correction systems try to correct the entire response (speaker + room) which results in many issues as a simple response curve does not accurately represent everything that occurs over several hundred milliseconds. The room treatment should be responsible for what happens to the sound beyond the direct sound from the speakers and most attempts to correct this digitally result in strange behavior. Having good off-axis behavior in a speaker is critical for this reason.
Correcting the response of the drivers in the speaker on their own affords many benefits which most speakers on the market do not have. Most speaker designers must aim to create a system that is relatively flat based on the response curves of the drivers they use. This limits their choices quite a bit and they must often sacrifice other aspects of the speaker’s performance. By using DSP correction, the response curve of each driver is fairly trivial which allows the drivers to be selected based on other characteristics. This includes distortion specs, dispersion characteristics. and any other anomalies. By prioritizing these aspects over a naturally flat response curve, an incredibly detailed speaker with extremely low distortion can be created. Control over the dispersion characteristics also creates many new possibilities such as the crossfire horns in my speakers which result in a sweetspot that’s wider than anything else around. More on that in another post.
This type of correction requires filtering beyond what typical IIR filters provide. The DEQX uses a combination of FIR and IIR filters to accomplish this. This allows for the phase and timing domain to be controlled separately from the frequency domain to correct timing anomalies in each driver.
The next aspect which the DEQX controls in my system are the crossovers. By using DSP crossovers, there is no filtering that happens after the conversion stage. Neither active nor passive crossovers. This eliminates the distortions caused by those circuits. Very few designers can create excellent passive crossovers and even then, they cause amplifiers to behave oddly. In most cases active crossovers will perform better than passive but also introduce distortions of their own. Both active and passive crossovers have severe limitations on the filter shapes which they can produce. On top of this, most designers will apply basic filter shapes without taking into account the response curves of the speaker drivers.
By using FIR filters, the DEQX is able to create perfect linear phase crossovers ranging from 48dB/octave to 300dB/octave. This results in no phase distortion which is not possible in other crossovers. The benefit of using steep crossovers (which also isn’t possible in other systems) is that there is little overlap between speaker drivers so the imaging is incredibly precise. The resulting crossovers are completely invisible.
The last area which DSP contributes to my system is in defining the low end of my speakers. Without DSP, my speakers would have a very strange frequency response which is by design. The tuning frequency is below where the drivers naturally roll off. By using DSP, the drivers are extended lower where they meet the tuning frequency of the cabinet. This wizardry adds roughly 10Hz getting the speakers down to 20Hz.
One last point to note is that any ported system should use a high pass filter below the tuning frequency. Without it, those frequencies will cause severe distortion without producing any output from the speakers. By adding a high pass filter, you’ll end up with a cleaner and tighter low end response which appears to extend lower. This is not possible with passive filters so must be done at least with an active filter but ideally with DSP.
The only real drawback to DSP when done correctly is the added expense and system complication. In my particular case, my left and right speakers alone require 6 channels of DACs and amps. The DSP does add latency, around 24ms in my case, so it’s not usable for most tracking but it’s not an issue for mixing and mastering where some plugins add several times that amount of latency.
If you have any questions, want to implement some DSP in your system, or still think that no DSP is better, get in touch.