SPL Auto Dynamic De-esser

Paul White makes a bid to trip up the SPL de‑esser, but even the phrase 'several sad somnambulists sincerely take tea tearing tickets till sunset' does little to ruffle its electronic feathers.

I have to admit that I've never been a strong advocate of conventional de‑essers — the cure somehow always seems worse than the symptoms. A conventional de‑esser is basically a compressor with an equaliser in the side chain. The equaliser is set to emphasise the offending sibilant, usually found in the 5‑8kHz range, so that the compressor becomes more sensitive to these frequencies, causing it to reduce the overall signal level momentarily whenever sibilant frequencies are detected. The problem is that sibilant frequencies don't always come along neatly separated from other bits of audio that you want to leave intact, so whenever a sibilant is detected, down comes the gain of the whole signal, whatever frequencies it comprises. The subjective outcome of this approach is that once you apply more than a very modest amount of de‑essing, the voice being processed seems to develop a most unnatural lisp! Why this has been tolerated for so long is a mystewee — it's a mystewee...

More sophisticated de‑essers have been available for some time, and they get around the problem described above by applying gain reduction only to the problem frequency band, leaving the rest of the spectrum unprocessed. SPL take this approach, but as you'd expect from SPL, some lateral thinking has gone into the design, to make the processor easier to use, and to keep audio side‑effects to a bare minimum.

Firstly, the SPL system doesn't just look for sibilance in a wide 'general sibilance' region of the spectrum. Instead, the user sets a front‑panel switch to Male or Female, in order to narrow down the search area. This centres the search in either the 6kHz or 7kHz areas respectively. In addition, the circuitry then automatically tracks sibilant frequencies in such a way that only the offending part of the spectrum is treated, effectively confining the processing to a much narrower frequency band. Instead of simply splitting the signal into different frequency bands and compressing the sibilance band, the SPL approach is to detect sibilant sounds, as above, then add these, out of phase, to the main signal, so that they cancel out. This neatly avoids the potential for phase and level errors introduced by band splitting.

Another unique feature of this de‑esser is SPL's automatic threshold‑adjustment system. With a conventional de‑esser, once you've set the compressor threshold, the input level must be kept reasonably constant, otherwise the amount of processing will vary with the level of the singer's performance. In real life, singers move in relation to the microphone, so to ensure more stable operation, the designers have devised an auto threshold system that automatically adjusts the de‑esser threshold according to how the input level changes. In theory, you could place a compressor before a conventional de‑esser to even out level changes, but in practice, the most natural‑sounding results are achieved by compressing after de‑essing.

The outcome of all this imaginative work on the part of the designers is that the user only has to set the Male/Female switch and then adjust the S‑Reduction knob to obtain up to 20dB of sibilance reduction.

The Rack, Please

Housed in a 1U‑high rack case, the SPL Auto Dynamic De‑esser features one of SPL's distinctive laser‑etched front panels, with a blue marbled finish. Perhaps this interesting effect makes up for the lack of controls to look at, because the control system is so beautifully simple that it's really quite boring.

The audio signal path is electronically balanced at a nominal +6dBu, and both TRS jack and XLR connectors are provided, as is a ground‑lift switch. This is a two‑channel unit where both channels operate independently; controls for each channel comprise just one knob and three buttons.

As I mentioned earlier, the de‑essing process is controlled by the S‑Reduction knob, and the Auto Threshold button activates the system that links the de‑essing threshold to the input signal level. Disengaging Auto Threshold leaves the threshold independent of input level, but for most applications, the unit works best with Auto Threshold left on. The Male/Female switch centres the de‑ess 'seek' zone at either 6kHz or 7kHz, but on occasions, you'll find the female setting works best with a specific male voice, and vice versa.

Bypassing the process is achieved using the button labelled Active; in bypass mode, the input signal is routed directly to the output, so that absolutely no noise or coloration is introduced. The same hard bypass comes in if there's either an interruption in the mains power supply, or a failure of the internal PSU. The audio signal path is exceedingly clean, with a frequency response extending from 10Hz to 100kHz (‑3dB), and when processing is taking place, the signal‑to‑noise ratio is quoted at ‑81dBu. In practice, this translates to 'extremely quiet': in all normal circumstances, the unwanted noise coming in via the mic or from the mic amp will exceed this figure by a considerable margin.

Brahms And Lisp

Testing this unit didn't take long — after all, the learning curve regarding the controls is so flat you could skate on it! With Auto Threshold switched on, turning up the S‑Reduction control to give four or five dB of 'ess reduction' is enough to make a considerable difference to the level of sibilance without unduly changing the overall character of the voice. What's more, the amount of de‑essing remains reasonably constant when the microphone is moved to simulate a typical performance or recording session. Using more de‑essing than strictly necessary leads to a slight dulling of the sound, but the side effects are significantly less serious than with a conventional de‑esser.

Switching Auto Threshold off doesn't reduce the effectiveness of the de‑esser, but the gain reduction meter does show that the amount of de‑essing taking place varies significantly as the mic is moved around. This is exactly as expected.

The Male and Female de‑essing frequency settings produce slightly different subjective results, and the sound of the remaining signal when processing heavily is also changed. Again, this is not unexpected, but it does indicate that you shouldn't take the male/female settings at their face value — try both and see which works best for any individual singer.

Summary

SPL have succeeded in taking the pain out of de‑essing, and although some slight timbral change is evident when heavy processing is being used, there's no sense of lisping or gain pumping, even at the most extreme settings. Using sensible settings, all that gets left out is the sibilance.

Most de‑essers produce such bad side effects that I'd rather change mic or mic position in an attempt to avoid having to de‑ess, but SPL's approach is eminently usable. What's more, the controls are so simple that even the most inexperienced engineer will find it virtually impossible to get into trouble. De‑essers aren't exciting, and this one isn't exactly cheap, but for the professional studio operator who has to be able to deal with all eventualities, including sibilant vocalists, it's a wonderful tool to have on hand.