There’s a very common attitude amongst people interested in audio that to a greater or lesser extent analogue equals good and digital equals bad. We all know this isn’t true but most of us to some degree see digital as a necessary evil in some way, something which at best does no harm but analogue has the potential to make things better by virtue of being analogue. Even as I type this I know it to be untrue but it’s still an emotional response I recognise even if I like to think I understand audio well enough not to actually believe it.
Nowhere do I see this more clearly in myself than my slightly begrudging attitude to DSP in speakers. I’ve used plenty of them and I know it works. The best implementations work what I can only describe as borderline miracles. It’s an acronym we casually throw around but what does it actually mean?
What Is DSP?
You probably know DSP stands for Digital Signal Processing. You’d be right in thinking that your DAW processes digital signals too but there is a difference in that DSP is performed on dedicated hardware which is optimised to perform repetitive, numerically intensive tasks as quickly as possible. The microprocessor in your computer is a more general-use device which can perform the same operations as a DSP but would do so less efficiently, consuming more power, generating more heat and probably incurring greater latency.
What Kind Of Tasks Are DSPs Used For In Speakers?
Signal processing can be performed in the analogue domain, hence specifying digital signal processing. Analogue capacitors, inductors and resistors can be combined in various ways to make very precise crossover filter networks which are of course a necessary part of two and three way speakers. Designing good analogue crossovers isn’t easy. Components affect each other in complex ways and the tolerances of analogue components vary, components with tight tolerances are more expensive and there is a practical limit to the steepness of the filters which can be achieved using analogue technology. A DSP can offer several tangible benefits straight away by way of repeatable performance. As a digital process the issues around component tolerances don’t apply and the cost burden of expensive components with tight tolerances disappear.
Higher order filters can be applied in the digital domain, filters as steep as 16th order (96dB/Octave) can be achieved, though the steepest filters in common use in crossovers typically remain at 24dB/Octave. However the advantages of DSP go further than just doing what analogue filters can do more accurately for less money, Using DSP it is possible to construct phase linear filters, something which isn’t possible in the analogue domain. When designing filters frequency and time are inter-related and using analogue components making the desired changes in amplitude at specific frequencies inevitably introduces phase shifts. In the digital domain it is possible to create linear phase filters at the expense of the time performance as this has to involve incurring a fixed amount of latency of between around 1 to 50ms. Small amounts of latency are possible but at the expense of low frequency detail, similar to what you’ll have seen If you’ve ever experimented with the window size setting in a spectrum analyser plug-in.
HEDD Lineariser - “DSP” Without Any DSP
The most common applications of DSP processing in monitor speakers is in the crossover design, particularly with the small but worthwhile improvements in performance which can be achieved using linear phase filters as a switchable option for use in playback only situations, when the latency they incur won’t be an issue.
An interesting example of just this kind of processing is the HEDD lineariser plug-in which performs exactly the kind of phase correction a linear phase filter would achieve but performs it natively, in the host computer using native processing from a general purpose microprocessor - so not using a DSP processor and not performing the crossover filtering, this is done using a traditional analogue crossover but HEDD have measured the phase errors introduced by these filters and created a plug-in which manipulates the phase into the exact opposite so that when played through the corresponding HEDD monitor the pre-manipulated phase emerges with a linear response.
Why do this? It’s an original take on an old problem. One benefit is that it saves on two conversion processes: one A/D and another D/A as the signal is digitised before entering the DSP and then converted back to analogue before being amplified.
EVE SC207 - Broad Brush Easy To Use DSP
An example of a monitor which uses DSP built into the hardware would be EVE Audio. DSP is an extremely powerful technology and whatever the problem, DSP probably has the solution but the trick is understanding the problem. EVE’s intention with their DSP is to present high quality processing which can tailor the sound of their monitors to a room in a familiar, analogue presentation without the users needing a knowledge of acoustics, just a pair of ears.
The DSP follows a 192kHz/24bit A/D converter and presents the user with control over a set of filters which allow the user to compensate for the influence of the speaker placement in the room and listening distance. These filters are often implemented in the form of dip switches on the back of each speaker, offering various dips and/or boosts at specific frequencies. EVE Audio provides access to their filters from the multi-purpose front panel LED-based adjustment knob. This sounds obvious but we’ve often been frustrated by the experience of adjusting the response of a speaker from the worst possible place to hear the effect - behind it!
Speakers placed too close to the walls often result in a build-up of low frequencies. Less than ideal positioning is sometimes an unavoidable reality in small home studios. Onboard DSP in the form of a low filter corrects this by dipping the low-end frequency range slightly to compensate. Rather than a simple one or two position dip switch, Eve Audio offers greater flexibility by providing attenuation below 300hZ in 0.5 dB steps. In the event your room modes are thinning out the low end, this range can also be boosted.
A high-shelf filter allows you to boost or attenuate frequencies above 3kHz in 0.5 dB steps. This feature is particularly useful when your listening position is less than ideal. The further you are from the monitors, the weaker the high frequencies. In such cases, the filter should be set to boost them. In desktop applications, distances can be as short as 1 m (3 ft), which results in increasing the high frequencies. To compensate for this effect, you should use the High-Shelf Filter to attenuate the high frequencies. If your room is dead sounding (lots of soundproofing material), a slight boost might be in order. Conversely, if you have a lot of reflective surfaces, a small roll off will help.
EVE Audio offers a third filter as well, the Desk Filter. It serves two functions. When turned down it acts as a narrow-band EQ set to either 170 Hz or 160 Hz (depending on the speaker model and size of the woofer). This cut is intended to compensate for the reflections emanating from your mixer or physical desk. When turned up, the Desk Filter also works as an EQ but this time set to 80Hz. This boost allows you to give more punch to the lower frequencies. Each of the three filters provides a boost of up to +3dB and a maximum attenuation of -5dB. all in 0.5 dB increments. And they can all be locked via the dip switches on the back.
Neumann KH80 DSP - Choose How Involved You Want To Get With Your DSP
These Neumann monitors are small in size but their use of DSP is interesting as it illustrates both the benefits and the potential pitfalls of DSP. I mentioned at the beginning of this piece that DSP offers greater flexibility in filter design than can be achieved in the analogue domain. Infinite Impulse Response filters share the same characteristics as analogue filters but can be much steeper. The KH80 DSP uses 48dB/Oct filters in the crossover with a linear phase response, something which isn’t possible without the use of DSP.
In the same way as the EVE monitors they offer easy to operate filters for basic correction to compensate for placement, boundary effects and desk reflections but they go further by offering more sophisticated “guesstimate” filtering based on the results of guided questions via an app about the room and the speakers placement in the room. This allows the monitors to be set up specifically for the room they are in without any measurements being involved.
A third option of direct manual control over the 8 filters per speaker allows detailed setup for those with suitable equipment and enough knowledge of acoustic measurement to measure the room accurately. Find out more about the KH80 DSP in our review article.
Moving Speaker Calibration From Computer To Monitor DSP
As we’re starting to see, DSP has the answer, the difficult bit is asking the question accurately enough. If you give enough information about your environment then given enough DSP processing power there is an awful lot DSP processing can do to help. However acoustic measurement is complicated and the quality of results is dictated by the quality of the measurements. There are a few very comprehensive solutions available which can be used to make acoustic measurements.
The Genelec GML system offers an extremely powerful set of proprietary tools for setting up Genelec DSP monitors but it is Sonarworks . who, through their software only approach have probably helped more people make useable room measurements than anyone else. Their user friendly software guides the user through making multiple measurements and calibrating monitors to their environment. However some people find the software only approach less convenient than having the processing built into the monitor speakers themselves.
IK Multimedia are soon to release a new monitor in their iLoud range called the MTM which boasts DSP for hosting their popular ARC 2 speaker calibration software.
IK includes their own ARC microphone in the package that you use to measure your listening position which in turn provides the DSP with a frequency response curve of your monitoring environment. This curve is then hosted on the DSP to deliver an ultra-flat frequency response. This system completely does away with having to host a speaker calibration plug-in within a DAW or host computer which is neat solution. The DSP in the monitor that hosts the correction will enable us to record, mix and monitor with speaker calibration active with no latency, fuss or stress about deactivating speaker calibration before bouncing sessions. The original iLoud Micro monitors are great sounding devices given their small footprint. The MTM build on the success of these by being a three way active monitor solution with ARC’s technology being built-in.
As speaker calibration has grown in popularity over the years it has become clear that moving the process of speaker calibration from the computer into the monitor via DSP is the future. IK Multimedia are not the only ones exploiting DSP in monitors. Earlier this year ADAM Audio announced their partnership with calibration specialists Sonarworks to develop a form of speaker correction in ADAM’s flagship S Series monitors.
All ADAM Audio’s S Series have powerful DSP chipsets built-in which enables their users to dial in a custom EQ curve via a piece of software to help their users combat pesky room issues. It’s works but the application isn’t really developed enough. We look forward to see and hear how the Sonarworks tech integrates with the ADAM monitors as the S Series sound mighty fine.
DAD AX32 With SPQ Card - DSP Muscle For Those Who Know How To Use It
The DAD SPQ card for the AX32 offers extremely powerful processing but offers no tools to help make measurements. As a professional solution it’s not unfair to assume such a unit will be set up by skilled operators. In fact in the kind of environment where an AX32 is likely to be used the monitoring environment is likely to be surround and quite possibly a Dolby Atmos system. To provide a DSP solution to calibrate a system with this many channels requires much more DSP and far more individual filters than would be necessary in a stereo monitoring system.
The SPQ card is an option for the AX32 (and the Avid MTRX) which provides up to 16 IIR filters per channel across 128 channels at 48KHz and can work to a maximum sample rate of 384KHz. It provides a total of 1024 filters to be allocated across the 128 channels as appropriate. This is a very serious DSP solution.
Kii Three - If You Can Afford It, Rather Than Compensate For Your Room You Can Ignore It!
The Kii Three is the perfect example of a speaker which rather than being enhanced through use of DSP is a speaker which couldn’t exist without DSP. Uncompromisingly modern and not cheap, these monitors do something different and they do it really well.
We’ve already differentiated between using DSP to optimise the response of a speaker. Using linear phase IIR filters and DSP too correct the frequency, phase and even correct timing between drivers DSP can improve the performance of monitor performance as heard independently of the environment in which it’s being used. DSP can also be used in conjunction with accurate acoustic measurements to calibrate thre response of a speaker so it sounds closer to flat even with the contribution of coloured reflected sound from the room. What the Kii Three does is control the directivity of the sound and minimise the amount of sound reaching those walls in the first place. By keeping the sound away from the walls the listener hears much more direct (uncoloured) sound in comparison to reflected sound. You can’t completely ignore the contribution of the room but the effect is very significant. This is why I described it a “jaw dropping” in my listening test.
The reason everyone doesn’t do this is because its difficult and it couldn’t happen at all without - -you guessed it - DSP. Controlling the direction in which sound radiates is relatively easy at high frequencies but becomes increasingly difficult the lower the frequency becomes. Deep bass is effectively omnidirectional. this is very inconvenient from an acoustic point of view because the biggest acoustic issues in rooms almost invariably occur at low frequencies. By using 6 drivers per speaker, each driven by its own amplifier and controlled by DSP which manipulates the phase and timing of the bass drivers the Kii Threes achieve a cardioid response across the full frequency range.
ATC - Do We Have To Use DSP At All?
The short answer is of course no. Taking as an example the ATC SCM25, this is a monitor which relies on old fashioned analogue engineering and delivers enviable performance which would satisfy the most demanding client. Building drivers as well as you can, putting them in the right cabinet and designing the best possible crossover results in exceptional performance. Put a speaker like this in a properly treated room and the results will speak for themselves.
The thing which probably should be said is that it isn’t an either/or choice. People talking this approach frequently add Sonarworks or a Trinnov to further improve performance. DSP can be cheaper, it can be better, it can certainly be more flexible, something to think about when you are pricing up acoustic treatment - will it be suitable in your next studio? Your DSP isn’t site specific. In the final analysis DSP can provide the answer, but only is you ask the right question of it and like the computer in the Hitchhiker’s Guide To The Galaxy, if you don’t ask the right question, don’t blame the technology if you don’t get the answer you were thinking of