DBX PROVOCAL

DBX PROVOCAL

Physical Modelling Recording Channel

 

In addition to the usual dynamics and EQ facilities, this voice channel from Dbx offers physical modelling effects in the digital domain, allowing it to simulate the performance of different types of mics and mic amps.

---

Paul White

Recording front-ends vary in complexity from simple mic preamps to complex channel strips with full dynamics and equalisation. Most are essentially analogue devices, albeit with a digital output option sometimes available, and typically you have the choice of solid-state or tube circuitry. The Dbx ProVocal channel strip is somewhat different, because, although its mic preamp section is necessarily analogue, the rest of the signal chain is digital.

Another interesting feature is that, in addition to compression, de-essing, gating and parametric EQ, it also offers mic modelling and preamp modelling, as well as a range of digital effects covering all the usual chorus/flange/delay and reverb treatments. Its outputs are available both on balanced analogue connectors and as a 24-bit S/PDIF digital signal (at 44.1kHz or 48kHz) on the usual phono connector. A degree of MIDI control is also provided, via the single rear-panel MIDI In socket. While most of the elements of the unit are fairly familiar, the modelling side of the product deserves a little more attention.

ProVocal Physical Modelling

Mic modelling is a contentious issue, as it purports to make one type of mic sound like another — clearly a process with inherent limitations. For example, while it might be reasonable to emulate the restricted high end of a dynamic microphone using a wide-bandwidth capacitor microphone as the source, trying to do the opposite is clearly a tall order. Furthermore, the characteristics of the source mic needs to be known reasonably accurately in order to derive an accurate transfer characteristic.

Dbx ProVocal£400 pros Lots of processing plus a digital output at a bargain price. Creatively useful mic modelling and effects. Good basic sound quality. Type IV conversion avoids digital clipping. cons Inadequate metering of levels and gain reduction. Inadequate parameter read-out and no patch naming. Most of the amp models are too distorted to be useful. summary This potentially powerful processor is let down by poor displays and the tendency to sound heavy-handed.

In the case of something like the Antares Microphone Modeller, the machine incorporates a database of many popular microphone types, whereas the Dbx ProVocal uses averaged data taken from several mics so that the user only has to choose from capacitor or dynamic source microphone types. Clearly any deviations of the source mic's actual frequency response from this notional average will affect the algorithm's effectiveness when it comes to predicting how the modelled mic would react to the same input sound source. In addition to modelling frequency response, a convincing emulation would also need to take into account phase shifts and distortions in both the source mic and the mic being modelled, but one aspect that can't be modelled is the off-axis response of a microphone, because that is the result of the physical attributes of the capsule, not the circuitry that follows it.

So, while it may be possible to produce a very close approximation where the source mic has lower distortion and a wider bandwidth than the modelled mic, and where the recording is taking place in conditions where reflected sound doesn't contribute in any significant way to the result, any real-life application of this technology can produce at best an approximation. Having said all that, if you approach mic modelling as being there to give you a choice of tonal characters from which to choose, rather than being a means of faithfully recreating the sound of a top-flight studio mic from something you found at the bottom of your gig bag, it can be a very useful technology.

Amplifier modelling has a far better chance of success, as the relevant parameters are measurable and, in the main, constant. While the main task of an amplifier is simply to make signals larger, many designs introduce non-linearities or tonal changes that add to the character of the sound, especially those utilising tube circuitry. If these idiosyncrasies can be modelled fairly accurately, then there's a good chance that the sound of a classic mic preamp can also be emulated. However, doing this properly requires that the input stage loading of the original mic amp is also recreated, which means using a variable-impedance input stage. This is technically difficult to do and, as far as I'm aware, it has not been attempted here.

Dynamics, EQ & Effects

The Dbx ProVocal is a 1U, single-channel processor that can accept mic, line or instrument inputs and generates a stereo output (necessary because of the effects) available on balanced XLR, balanced jack or as S/PDIF. The maximum analogue output level is +18dBu. An electronically balanced mic preamp provides switchable phantom power and a 20dB pad, along with simple three-LED metering of level. Both XLR and high-impedance instrument jack (450k(omega)) inputs are to be found on the front panel with a Line switch that engages a rear-panel balanced line input jack instead of the front-panel instrument jack.

After that point, the signal is digitised using Dbx Type IV conversion, an ingenious proprietary system that provides linear conversion up to a level of 4dB from clipping, after which a nonlinear (logarithmic) conversion system is employed to implement a soft saturation characteristic that extends the dynamic range enormously and makes the converters virtually impossible to overload. I've used this system before and like it very much, as it allows you to work close to the full-scale level without ever having to worry about clipping and the unrecoverable loss of high-frequency information (from the top of the clipped waveform) that entails — not to mention the evil-sounding distortion!

As with any digital system, there are far more parameters to control than there is space for knobs, so a simple matrix system has been employed where five knobs address six rows of parameters, all clearly printed on the front panel. These are regular pots, not rotary encoders, so you need to turn them through the stored value position in order to get them to pick up the parameter for adjustment. A two-digit LED display shows the number of the current program (40 factory and 40 user) and also serves as a parameter display when editing. Two buttons control the patch scrolling or can be pushed simultaneously to bypass all effects, while another button steps through the vertical axis of the effects matrix. That leaves just a Store button, for saving programs, and a power switch.

Most of the parameter rows are logically grouped. The mic model type, amp model type, drive amount and gain are controlled by the first row, along with the low cut frequency. The second bank is given over to a three-band EQ with shelving extremes and a parametric middle section, while row three deals with the compressor, row four addresses the gate and de-esser and the remaining two rows look after the effects.

No information is given concerning the ten mic amp settings other than their number, with the first setting being flat and the remaining nine getting progressively more tube-like. The Drive control adjusts the input drive to the amp model to simulate tube distortion. More information is provided for the mic modelling, where ten options are available for both dynamic and capacitor sources, if you count the flat option at setting one. Two of the models are special effects for telephone and megaphone simulation, with the rest modelled after distinctive models including the Shure Green Bullet harmonica mic and the famous RCA ribbon. The large-diaphragm models are based on the characteristics of popular Shure, AKG and Audio Technica mics, while the small-diaphragm model is based on an Oktava pencil mic.

In order to keep operation of the unit simple, the effects have user parameter sets more akin to those found on effects pedals rather than studio rack processors. For example, the reverb can be selected from ten types, after which only the mix and decay time may be adjusted, in this case by a single control that brings in more reverb level as you lengthen the decay time. Similarly, the delay (1380ms maximum) offers only time, feedback and mix, while the modulation effects offer type, depth, speed and mix. Although delay and reverb may be used together, you have to choose between chorus and flanging. There's a MIDI control facility, where the user can select one of ten variable parameters within the patch to be controlled by an outside MIDI control source.