The Path of Least Resistance

Sep 1, 1999 12:00 PM, Gino Robair

There is no question that direct recording with good gear will give you the cleanest sound. And no instrument reflects the efficiency of direct recording more than the human voice. However, the average vocal signal takes a trip through various gain stages in the mixer's channel strip, through the send/returns to effects (such as compression and EQ), and down the mix bus before hitting the recorder. Electrically speaking, that adds up to quite a distance and can translate into added noise and signal degradation.

For avoiding such noise and degradation while still getting the effects essential to recording vocals, there are handy devices on the market that reduce the number of parts the signal goes through and thus minimize the length of the path from input to output. These devices are often referred to as channel strips and voice processors.

ONE VOICE, ONE CHANNEL The terms channel strip and voice processor are often used interchangeably. That's because both kinds of units have a similar purpose: to get the signal to the recorder as directly as possible. But the two types also have significant differences.

Stand-alone channel strips are designed to take the place of a mixer's channel strip. The stand-alone has the same features as the channel strip on a mixer: mic preamp, line-level input, shelving filters, and parametric EQ. Besides the shorter signal path, the big difference is that the stand-alone version contains better components than those in the average personal-studio mixer. When you spend $600 per channel, you get components comparable to those in a high-end mixing desk.

Voice processors are also designed for direct recording and contain the same features as a channel strip, but with a few additions. Engineers regularly use dynamics processors when recording vocals because of the voice's wide dynamic range. That means you will want a compressor, limiter, expander, gate, and de-esser in the unit. Of course, these effects work well on other instruments, too, such as electric guitar, bass, and keyboards.

The layout of a voice processor should be as intuitive as that of a channel strip. In a mixer, the signal generally flows from top to bottom. The signal flow in a voice processor is similar, except that you lay the unit on its side so that the signal runs from left to right, entering the input stage at the far left, going through the various effects stages, and ending up at the output on the far right.

Don't limit yourself by using voice processors only for tracking. These units can prove equally beneficial during mixdown or for fattening prerecorded tracks. For this reason, voice processors have a variety of input and output options to maximize their usefulness in the studio.

CHOICES AND MORE CHOICES Manufacturers all have different ideas as to what their direct-recording device should include and how the features should be organized. Because of the enormous number of possibilities, I have limited myself to discussing a handful of products for the purpose of comparison.

Voice processors can range in price from around $500 to $4,000. I decided to look at single-channel units priced between $630 and $850 that combine a mic preamp, compressor/limiter, expander/ gate, and EQ. Within these parameters, I looked at a host of other useful features (for example, tube emulation, digital converters, and independently accessible effects), and chose a collection of processors that complemented, rather than fully duplicated, each other.

Besides the five that I have chosen to cover here, there are many other voice processors in this $630-to- $850 range. Therefore, as I discuss specific features, I will also briefly point to some of these others as examples. But before we examine our five contenders, let's review the various features that make up a voice processor.

INS AND OUTS When you're looking for the ideal direct-recording solution that is also useful for mixing, the more I/O possibilities you have, the better. In addition to a mic input, there should be line- and instrument-level inputs. The ability to tap in and out of the individual effects is a plus. At the other end, having balanced (+4 dBV, 11/44-inch TRS or XLR) and unbalanced (-10 dBu, 11/44-inch TS) line-level outputs is essential. Let's take a closer look at each section.

Mic preamp. The microphone preamp is one of the most important parts of a voice processor because it sets the stage (literally) for your sound. In order to record with the best signal-to-noise ratio possible, the output level of the microphone (-30 to -60 dBu) should be brought up to line level (-10 dBu or +4 dBV).

Usually, this change in level will color the sound somewhat, and in many cases that's good. One joy of engineering is tone sculpting, and the mic preamp can have as much to do with this sculpting as the microphone itself. For example, a tube preamp will have a different character than a solid-state unit. In addition, Class A circuitry (solid-state or tube) will give you superior sound and performance but will often cost a little extra.

Ultimately, the degree of transparency or coloration that you want the preamp to impart depends on your personal taste and the demands of the music. Other considerations include the recording medium that you're using. Engineers with tapeless studios may want a preamp that adds "tube warmth" or can emulate the effects of tape saturation. On the other hand, users of analog tape may want a transparent sound, with as little coloration as possible. The call is a completely subjective one. What sounds harsh and buzzy to one engineer may sound warm and fuzzy to another.

Another consideration is how your collection of mics will sound through the preamp: invariably, some mics will sound better than others. There are no rules that say you can't run a cheap mic through an expensive preamp or vice versa-which preamp you choose should be determined by whatever works best for the style of music that you're recording.

Examples of voice processors with tube preamps include A.R.T.'s Pro Channel ($799), Manley's Voxbox ($4,000), and Avalon Design's VT-737SP ($2,295). Solid-state designs include the Focusrite Voicebox MKII ($1,345) and the Rane VP12 ($599). The Manley, Avalon, and Focusrite are more expensive because they use Class A circuitry.

A mic preamp can include a couple of features that make life easier. First, it should have +48 volt phantom power so that you can power the mics that need it. A phase-reversal switch is also useful for changing the phase of the signal by 180 degrees. In addition, preamps often have a -10 or -20 dB pad to attenuate high- output mics and a low-cut filter to remove low-frequency energy such as rumble. Most voice processors on the market include many or all of these features.

Instrument input/DI. If you have ever recorded electric guitar or bass, you know how handy it is to have a compressor, gate, and EQ at your disposal. Having access to a tube stage is another plus. You'll probably want to plug your instrument into your voice processor and take advantage of these effects. Fortunately, many of these units have a 11/44-inch, instrument-level input jack. The instrument input (or DI) raises the signal from instrument level (-30 to -20 dB) to line level. Look for processors with convenient 11/44-inch front-panel jacks.

Line-level input and output. In addition to using a voice processor on mics and guitars, you can use it on keyboards, sound modules, and prerecorded tracks during mixdown. Manufacturers often provide both +4 dBV balanced and -10 dBu unbalanced line-level inputs to facilitate this type of use.

Voice processors also typically have a -10 dBu and a +4 dBV output. Depending on your studio situation, having the option of both output levels means that the device will have upward compatibility if you're moving from so-called semipro gear (which typically operates at -10) to professional gear (operating at +4).

Compressor/limiter. The dynamic range of the voice goes beyond what the average recorder can handle, so engineers routinely use compression when tracking vocals. The type of compressor and how well it operates can impact how the signal sounds after the fact. Consequently, take great care when choosing and using a compressor.

With a dedicated compressor you will usually have control over threshold level, attack and release times, ratio selection, and make-up gain, as well as the choice between hard or soft knee (that is, how suddenly compression begins).

Limiting is at the extreme end of compression, when the ratio is 10:1 and beyond. Limiting effectively inhibits any increase in output level no matter how far the signal goes above the threshold.

Compressors and limiters have three common designs: photo-optical, VCA, and variable mu. Photo-optical compressors (also called photoelectric or just plain "opto") differ in several ways from those compressors governed by a voltage-controlled amplifier (VCA). Opto compressors use a photosensitive resistor controlled by light-emitting diodes that are triggered by the input signal. Opto units compress over a narrower range (with less extreme compression ratios) and impart a livelier sound to the signal. Because of their simple user interface, they lend themselves to tweaking by ear. Examples of photo-optical compressors include A.R.T.'s Tube Channel ($499), Avalon Design's VT-737SP, and Focusrite's Platinum VoiceMaster ($749).

VCA units, on the other hand, are far more predictable, are easier to use, have greater ratios, and allow for more flexibility and precision with attack and release times. Voice processors with VCA compressors include the MindPrint En-Voice ($749), PreSonus VXP ($700), and Focusrite Red 7 ($2,995). The En- Voice compressor includes a tube saturation stage.

Variable mu compressors (mu in this case means "gain") use a vacuum tube as a variable resistor to control the signal level. Manley uses a combination of variable mu and optical compression/limiting in its Voxbox. The A.R.T. Pro Channel gives you the choice of either opto/tube or variable-mu gain control.

Expanders and gates. Downward expansion and gating are processes used to reduce noise. Expanders create a greater dynamic range by progressively attenuating sounds that fall below a specified threshold; that is, they make the quiet sounds quieter. You can think of an expander as the opposite of a compressor, which brings up the level of the quieter sounds.

A full-featured expander gives you control over the threshold where the processing begins, the ratio of expansion, and the speed of the processing. Expanders are used primarily for eliminating extraneous sounds such as bleed from headphones, lip smacks, and clothing noises. Usually used during mixing rather than recording, the expander is often placed after the compressor to reduce the low-level noise that compression brings out. Many voice processors, however, locate the expander before the compressor, giving you the opportunity to reduce noise before the compression stage.

At the extreme end of downward expansion is gating. Rather than simply reducing the signal level when it drops below the threshold, a gate shuts the level off completely. Gating is used as much to shape the envelope of signals as to remove unwanted noise. For example, gates are used to fatten up drum sounds by quickly attenuating the instrument's natural decay. You can also use a gate to keep the buzz of an amp from passing through the unit when a guitarist isn't playing. Typical gate parameters include threshold and rate.

EQ and filters. Direct control over the timbral aspects of a signal is always welcome. Whether you use EQ to enhance the signal, remove unwanted anomalies, or repair damage caused by compression, you will want a band or two of parametric EQ as well as high- and low-shelving filters.

A parametric EQ lets you control three essential parameters: the exact frequency that you want to affect; the bandwidth, or Q, which is the range of frequencies around the center frequency; and volume for cutting or boosting.

Shelving filters allow you to modify the extreme ends of the signal and are usually found in the input section of voice processors. The most common is the highpass filter (also called a low-cut filter) which attenuates everything below a specific frequency. This is especially useful for removing rumble and mic- handling noise. Because the human voice doesn't go below 80 Hz, a typical cutoff frequency for a highpass filter is 75 Hz. Some units give you variable control over the cut frequency, sometimes ranging from 15 to 320 Hz. At the other end of the spectrum is the lowpass (or high-cut) filter, which affects everything above a specific high frequency, such as hiss.

For those of us with a limited number of mics, parametric EQ supplies the tools with which to expand the sonic palette. EQ is also a must when tracking instruments such as the electric guitar. Ranges in which the guitar can often use help include the lower-mid range, around 500 Hz; between 3 to 6 kHz for added bite; and from 8 to 10 kHz for sparkle. Unlike a graphic EQ, a parametric EQ enables you to locate the exact frequencies that need attention.

De-esser. A de-esser provides a form of frequency-dependent compression used to remove sibilance when recording vocals. Sibilants, such as s and sh sounds, have a high-frequency concentration between 3 and 8 kHz. Traditional de-essing is done by splitting the vocal signal, sending one side through the compressor's audio input, and the other side to an EQ patched into the compressor's sidechain input. The exact sibilant frequencies are boosted on the EQ, which makes the compressor more sensitive to them. The EQ-enhanced signal causes the compressor to attenuate the sibilants in the direct signal. Meanwhile, the signal going into the sidechain is not heard.

A de-esser increases the usability of a voice processor. However, a poorly implemented de-esser can sound more like a lowpass filter than a frequency-dependent compressor.

Insert and sidechain. As with a mixing console, an insert lets you introduce an additional outboard processor into the signal path. It also gives you an extra output point before the main output stage.

The sidechain input, a common item on dedicated dynamics processors, lets you control the compressor with an external signal. This is useful for ducking and frequency-dependent compression.

THE CONTENDERS In designing a product such as a voice processor, manufacturers make presumptions about how recording engineers like to work and the kinds of features that they need most. For the sake of contrast and comparison, I have selected five units that have as many of the above features as possible, with just as many implementations.

dbx 1086. The dbx 1086 Mic Pre Processor ($750) stands out in the crowd in several ways. The most interesting for the personal-studio owner is that the mic preamp and the dynamics processor can be used independently. This allows you to simultaneously track a vocal while processing another signal. The other great feature is that the 1086 has room for an optional A/D converter, the dbx 504X. However, the 1086 is the only unit of the group that lacks a true parametric EQ section (see Fig. 1). The 1086's preamp section has an XLR input and an XLR and 11/44-inch TRS output. The dynamics section has XLR and 11/44-inch TRS line-level inputs and outputs. Both output sections have a switch allowing you to choose between -10 dBV and +4 dBu output levels (see Fig. 2).

The preamp section has a variable-frequency low-cut filter (30 to 300 Hz) and a 2-band additive filter called Detail. When switched in, the Low control simultaneously boosts 125 Hz and cuts 400 Hz at a 2:1 ratio. The High knob adds up to 15 dB of the "air band" above 10 kHz. Having a variable-frequency Low Cut next to the Detail section baffled me initially. But I could imagine a scenario in which you would want to boost 125 Hz while cutting 70 Hz at the same time, and this unit will give you that option.

The 1086 has a single bypass switch for the entire dynamics section. However, the controls for the expander/gate, compressor/limiter, and de-esser each have an off position that effectively removes them from the signal chain. All of the buttons on the 1086 are backlit so that you can easily see their status, and the rotary controls use 40-position stepped, rather than continuously variable, pots.

The expander/gate has variable threshold and ratio controls, the compressor has threshold, ratio, and output-gain controls, and the de-esser lets you set threshold and frequency. The VCA compressor includes an OverEasy switch for soft- and hard-knee processing, and a Slow button that changes the speed of the compressor. The 1086 also includes the dbx PeakStopPlus two-stage limiter to keep the output from overloading the recorder input.

Drawmer MX60. Officially called the Front End One, the MX60 ($629) has all the traits of a well-equipped voice processor (see Fig. 3). The rear panel includes +4 dBu balanced and -10 dBV unbalanced 11/44-inch line inputs, a mic input, and an insert jack. An instrument-level input is strategically placed on the front panel. The MX60 can simultaneously handle balanced and unbalanced line-level inputs, and balanced and unbalanced line-level outputs. Together, these features allow you to use the MX60 for level conversion (see Fig. 4).

The dynamics section of the MX60 includes a gate, compressor/limiter, and de-esser. The design of the MX60's VCA compressor is based on the DL241 and MX30. The three bands of EQ include a fully parametric mid band (with frequency, bandwidth, and cut/boost controls), as well as high- and low-shelving filters. Next is Tubesound, which has three separate adjustable bands of saturation: Lo covers 350 Hz and below; Mid handles 350 Hz to 2 kHz; and Hi is 2 kHz on up. Each band has a Drive control that ranges from 0 to 11.

The MX60 has a fixed-threshold (+20 dBu) prefade limiter before the master output fader. The limiter isn't user selectable-soft limiting begins automatically at +6 dBu-so you really have to watch how you manage the various gain stages to avoid unwanted limiting.

Finally, the MX60 has no power switch. To avoid sending damaging current spikes to your mics when you turn on your power strip or line conditioner, the unit's phantom power comes up and decays slowly.

Focusrite Platinum VoiceMaster. The Platinum series is Focusrite's foray into the price realm of the average personal studio. While products in their Red series cost upward of $2,500, the units in the Platinum line are priced well below $1,000 and contain many well-thought-out features based on the more expensive models.

For example, the Platinum VoiceMaster ($749) contains a Class A discrete transistor mic preamp with a frequency response of 10 Hz to 200 kHz (with -1 dB variance), which exceeds the human hearing range. Another notable feature is an opto compressor that includes a Treble control for reintroducing high frequencies into heavily compressed signals (see Fig. 5). The threshold and release times have variable controls, while speed and ratio are set with two-position buttons: Attack time is set using the Fast button, while Hard Ratio gives you a choice between high (6:1) and low (2:1) ratios. In addition, the VoiceMaster has an opto de-esser as well as a tunable saturation stage. An expander/gate and EQ section round out the feature set.

VoiceMaster's back panel includes mic- and line-level inputs, an insert jack, and an XLR output that lets you take a signal out before it gets to the de-esser (see Fig. 6).

HHB Radius 40. HHB distributes TL Audio's Ivory Series 5051 Valve Processor in North America under the name Radius 40 Tube Voice Processor. (See the review of the 5051 in the December 1998 issue of EM.) Designed primarily for vocals, the Radius 40 ($749) is the only unit in this group without a built-in de-esser. A sidechain jack is included on the back panel so that you can de-ess the old-fashioned way.

Inside the Radius 40 are three tubes. The first is used for the input stage as part of a solid-state/tube hybrid circuit. The second and third are used in the compressor and 4-band parametric EQ stages. As the levels through the tube stages increase, the amber Drive LED illuminates. The red Peak LED signals that there is less than 5 dB of headroom left.

The input section includes a front-panel 11/44-inch jack that can accept instrument-level and line-level inputs, and a switchable 90 Hz low-cut filter (see Fig. 7). The back panel has a mic input, 11/44-inch unbalanced and XLR line-level inputs, the sidechain insert, a link jack for synchronizing the VCAs of two Radius 40s, an input-level switch, and XLR and 11/44-inch unbalanced outputs (see Fig. 8). Interestingly, the Radius 40 lacks a phase switch, which seems an odd omission for a device that can be linked into a stereo configuration.

The Radius 40 has a proprietary solid-state compressor called a transconductance amplifier, with four fixed attack and release times. Threshold, ratio, and gain have continuously variable controls. The attack ranges from 0.5 to 40 milliseconds, and the release times are from 40 ms to 4 seconds. The compressor can be switched out of the signal path using a front-panel button.

Each of the EQ bands has four fixed bandwidths and a variable level control. The two mid bands have fairly wide bandwidths so that frequencies in adjacent bands overlap. The EQ section follows the compressor in the signal path but can be easily switched ahead of the compressor using a button on the front panel.

LA Audio PS-1. The most expensive voice processor of the bunch, the PS-1 ($850) has a mic preamp (with high- and low-cut filters), a full-featured compressor (including variable control over attack and release times), a parametric EQ with two variable mid bands, an expander, and a de-esser (see Fig. 9). In addition, the PS-1 can be fitted with an optional A/D converter.

The back panel gives you a good deal of flexibility by having separate inputs and outputs on the EQ and dynamics sections (see Fig. 10). The input section has mic, line, and DI inputs and a 11/44-inch output. The dynamics processor has a line-level input and output as well as stereo-link and sidechain jacks. The EQ has line-level input and output, and the Output stage has a 11/44-inch input next to the XLR output. All this I/O flexibility means that you can shorten the path to the recorder by using the preamp's 11/44-inch line output, use the different effects independently, or reorder the effects in the signal chain.

SONIC FINDINGS The best way to test a voice processor is on a voice. I asked singer/songwriter/ guitarist Jill Garellick of Cactus Motel to be my test subject because of her dynamic singing and strumming style. I also recruited engineer Myles Boisen for his ears and gear. Garellick's guitar and voice were recorded separately using a matched pair of mics so that I could use two voice processors per take. I recorded the guitar using a factory- matched pair of Oktava MC012 microphones and tracked the voice with a matched pair of '70s-era Neumann U 87s. Each mic went through a single processor and then directly to tape.

To further test the behavior of the units, I ran a few different instruments through them, including guitar, bass, theremin, keyboards, and drum machine. I also routed tracks from tapes (music and spoken word) through them to see how each device reacted to various tracking and mixing conditions.

Pump up the volume. Each of the mic preamps in the collection has phantom power and a low-cut switch, and all but the Radius 40 have phase reversal. Other than features, what really sets these preamps apart from each other is how they sound; when heard side by side the differences are remarkable.

The combination tube/solid-state design of the Radius 40 mic preamp gives it more coloration than all but one of the other preamps, but with an added throatiness. Boisen described its sound as "warmer or murkier, depending on your taste," a quality that is no doubt due to the tube influence. The Radius 40 does, however, have a pronounced upper midrange that helps counteract its preamp's slight muddiness.

Heard on its own, the Radius 40 preamp is noticeably fuzzy. Even in the clean settings, the contours of the voice sounded like they were wrapped in gauze. Once the Radius 40 tracks were placed in the mix, however, they sounded great. The Radius 40 guitar track blended well with the vocals tracked through the other processors, and vice versa. With that in mind, I'd be less conservative with the degree of processing next time I use the preamp. Although it doesn't have as much "clean" headroom as the other preamps, you can push the overall level much further.

In contrast to the Radius 40, the PS-1's preamp had a smoother high end and an impressive clarity. The high harmonics of the guitar really came through. As far as tonal coloration, its preamp was neither the most nor the least transparent of the five units. Compared with the MX60 and the VoiceMaster, I detected a slight buzziness around the voice, similar to that of the Radius 40 but far less prominent. And the PS-1 preamp seemed to have quite a bit more headroom.

The preamp in the dbx 1086 was a little less satisfying to use. The stepped input gain was problematic at times: there were drastic changes with each step, and sometimes the level I wanted was between the notches. Adjustments to the level often affected the behavior of the compressor, making it more difficult to predict the response. And similar to the Radius 40, the 1086 had the least amount of headroom and required a greater input gain to get a level on tape that matched the PS-1, VoiceMaster, and MX60.

The 1086 was the least transparent preamp of the bunch. Hearing it on its own, I could detect a little coloration. In side-by-side comparisons, however, it sounded somewhat two-dimensional, and the coloration was far more pronounced. But coloration is not necessarily a bad thing. EM associate editor Brian Knave used the 1086 preamp to smooth out an otherwise harsh vocal sound. One might say that the 1086's preamp has a "soft focus" effect on the voice.

On the other end of the spectrum is the preamp in the MX60. Both the voice and the acoustic guitar sounded fantastic through it. Boisen noted a slight boost to the low mids, which added a bit of lumpiness to the sound. You have to mind the MX60's input level carefully, otherwise the prefade limiter starts working on the sound. If you want to avoid that stage altogether, take the +4 dBu signal out of the balanced 11/44-inch insert jack on the back panel. The MX60 Class A preamp includes a phase reverse switch, a 100 Hz low-cut filter, a Brightness button, and a -20 dB pad.

Of the five mic preamps, the VoiceMaster is the most transparent. As the other Class A preamp in the collection, it's easy to see why. The high end is crisp and clear, and the frequencies are more evenly distributed than in the other preamps. I compared the VoiceMaster to the Focusrite Green Series Dual Mic Pre, and the sound was remarkably similar.

Total transparency may not always work for a session, so it's nice to have the ability to gradually color the sound to taste. Both the VoiceMaster and the MX60 give you a tunable saturation stage that, on the voice, sounded convincing in very small doses. And while we're discussing saturation, coloration, and buzziness, let's see what happens when we run line- and instrument-level instruments through these devices.

Instrumental connections. Because TL Audio/HHB's and Drawmer's designers were nice enough to put instrument jacks on the front panels, I'll begin with the Radius 40 and the MX60. The 11/44-inch jack on the front panel of the Radius 40 can handle just about any signal. That's great, because going behind your rack each time you want to plug in an instrument is inconvenient. Keep in mind that it's an unbalanced input: if your keyboards are across the room and you need the RFI protection of a balanced line, you may end up going behind your rack after all. Be sure to have a TRS-to-XLR cable as well because the Radius 40's balanced input uses an XLR jack.

Like the Radius 40, the MX60 has an unbalanced 11/44-inch input on the face, though it is meant for instrument-level signals only. The back panel has +4 dBu balanced and -10 dBV unbalanced 11/44-inch input jacks, and you can use both of them, as well as the balanced and unbalanced outputs, at the same time.

In its back-panel input section, the PS-1 has jacks for line- and instrument-level signals, or you can go directly into the dynamics processor, equalizer, and output section if you like. The 1086 and the VoiceMaster do not have a built in DI. The only line-level input on the 1086 goes directly into the dynamics processor.

Dynamic differences. There are bound to be compromises when you jam a multitude of effects into one box priced under $850. This is most noticeable in the feature sets of the dynamics processors. As far as getting a full-featured compressor, the PS-1 and Radius 40 come the closest. They both offer control over attack and release times, as well as ratio, threshold, and make-up gain.

Besides having variable threshold (-30 to +20 dB) and ratio (1:1 to 20:1), the PS-1 is the only one of the bunch with continuously variable attack and release controls. In addition, it has a Gain knob with 20 dB cut or boost, as well as a hard/soft-knee button and a bypass switch. The PS-1 compressor can be used independently of the other stages, and it has a link function for stereo use.

However, the PS-1 compressor is more challenging to use than the other compressors in this group. The controls are very touchy, and the slightest movement noticeably affects the sound. And to get the meter bar to match what I was hearing, I had to push the input gain more than with the other units.

On the other hand, the dbx meters tell you the story right away. Although the 1086 has no release-time control, it is smoother sounding and more musical than the PS-1. You adjust attack and release times automatically on the 1086. By engaging the Slow button, you can extend the attack and release times for instrumental applications. But having so little control over the speed made the 1086 compressor challenging to use; getting the right setting was sometimes difficult due to the stepped controls.

Setting up the VoiceMaster's opto compressor, by contrast, was quick and easy. It has only two ratio settings to choose from-Soft Ratio (2:1) is intended for vocals and Hard Ratio (6:1) for instruments. However, the soft setting was perfect for acoustic guitar as well as voice. And having a variable release time helped in getting the right sound for electric guitar and bass tracks. The Treble feature proved useful in reinstating high frequencies postcompression.

The VCA compressor in the MX60 was a tad smoother than the VoiceMaster's opto compressor. Although the attack and release times are automatically controlled, I had no trouble getting a natural and musical sound. Besides a variable threshold control, the MX60 has a continuously variable ratio. The MX60's compressor cannot be linked for stereo use.

The Radius 40 has one of the smoothest, most gentle compressors of the group. It was easy to dial in a setting on both the guitar and voice, using medium-slow speeds and a ratio of about 8:1. Drum samples sounded especially beefy through this compressor. And having the option of switching the EQ in front of the compressor helped even out the compressor's response to bass-heavy signals.

Mind expansion. All but one of the voice processors in this group place the expander/gate before the compressor. The Radius 40's expander/gate is at the very end of the signal chain and is by far the simplest of the bunch. It is controlled by a single knob with a range of off/-50 to -20 dB. This expander/ gate acts as a gate in the higher settings (fully clockwise, between -30 and -20 dB). However, this results in envelopes getting sped up and sharp transients being chopped-so much so that, when set at -25, it gave drum samples a reverse-sounding attack.

Although the Radius 40's expander/ gate control comes after the master output control in the signal chain, the knob is located next to the make-up gain in the middle of the front panel. This clued me in about how to use it. When I really pushed the input and make-up gain stages, I could use the expander/gate to keep the noise of the twice-boosted signal from coming through. Still, I missed having separate control over the speed of the gate. Other than this one application, I had a difficult time using this particular expander/gate successfully.

Compared with the Radius 40, the 1086's expander/gate is not only easy to set up but sounds fantastic. Threshold and ratio are your only controls, and with the 1086, that's enough to shape percussion and keep noise to a minimum. The threshold range is off/-80 to +15 dB, and the ratios span 1:2.1 to 1:8 (although they are mistakenly printed as 1.2:1 and 8:1 on the front panel). At times, the gate seemed a little touchy, probably due to its automated program sensitivity.

LA Audio refers to the PS-1's expander as Noise Reduction. This may seem slightly confusing at first, but as a downward expander, it delivers on its promise. The Noise Reduction function has a preset ratio equivalent to 3:1, a threshold control that ranges from -70 to 0 dB, and a two-position, switchable release time (fast or slow). The attack is preset with a soft-knee envelope that sounded natural on everything I ran through it. An amber light lets you know when the signal is above threshold. The back-panel link connector allows you to coordinate the expanders of two PS-1s for stereo operation.

The PS-1's knobs are highly sensitive, requiring very subtle movements to get the right timing. However, I had no trouble dialing in a natural decay each time. As with the PS-11's de-esser and compressor, you can use the expander independently of the EQ and input section. The three dynamics processors share the same I/O, so you can't plug into each of them independently.

The gate Drawmer put in the MX60 is actually more of a downward expander. The MX60's gate has two speeds (fast and slow) and a variable threshold (off to +20 dB). As with the other gates, changes in threshold will vary the speed: when you dial toward a higher threshold the gate moves quickly, and lower threshold settings (the MX60 descends to -70 dB) slow the gate down.

It was often difficult to make the MX60's gate sound transparent. However, when I did find the right setting for an application, it worked well. The combination of a low threshold with the slow speed setting was perfect for fattening a noisy drum machine. In this particular instance, the threshold control made it fast and easy to find a decay that sounded musical.

The VoiceMaster's Noise Reducing Expander includes a switch to choose either gating or expansion. The variable controls include Threshold (-40 to +10 dB) and Depth (0 to Full). A 4-LED bar graph indicates the amount of reduction being applied to the signal. The VoiceMaster's expander worked well with vocal tracks (both spoken and sung). But when the gate was engaged, the speed was so fast that it was difficult to find a good setting using only threshold and depth controls. This particular gate could use a speed control.

EQ review. Like dynamics processing, a parametric equalizer is handy for both tracking and mixing. The most useful parameters in the midrange frequencies are adjustable bandwidth, sweepable frequency range, and cut/boost control.

The PS-1 possesses the most dramatic-sounding EQ of the bunch, with two full-featured mid bands in addition to high and low shelving. Each of the PS-1's mid bands gives you control over a range spanning 60 Hz to 20 kHz. The MX60 has a similarly wide frequency range of 150 Hz to 16 kHz. However, this range is packed into a single parametric band with shelving filters on either side.

The EQ in both of these units worked so well that I wanted to use them on everything! And it was easy to zero in on the frequency I needed. These EQs are so accurate that you don't need much boosting or cutting to hear a change in the sound.

The EQ on the Radius 40 was far less dramatic, but no less musical than the PS-1 or MX60. The combination of wide bandwidths and overlapping frequencies made this EQ particularly useful. Having a tube stage in the EQ is also a plus if you want to add extra coloration.

Although the 1086 has no EQ section, the preamp has a variable-frequency low-cut filter. I didn't find the dual-band Detail feature particularly useful on voice or acoustic guitar, though it may prove useful with other miked instruments. I was disappointed that the 1086 had no line-level input before the dynamics processor: I suspect that Detail might work well on thin-sounding electronic instruments, but there's no way to try it.

The voice-optimized EQ on the VoiceMaster has five controls, cryptically named Warmth, Tuning, Presence, Absence, and Breath. This section worked well once I figured out how each of the controls works. Breath is a 10 kHz shelving filter for cutting or boosting the "air" frequencies. Tuning and Warmth, combined, are like a low-frequency parametric EQ. Tuning determines the center frequency (between 120 and 600 Hz) that Warmth cuts or boosts. Presence is a fixed bandwidth cut/boost control with a peak at 1.5 kHz. The Absence button has a center frequency of 4.5 kHz and is intended to further attenuate the midrange by 6 dB. The area around 4.5 kHz is where the voice can sound harsh.

On the units that have equalizers, the EQ is placed after the compressor in the signal chain. Sometimes, however, you may want the equalizer before the compressor. For example, if a specific frequency is causing the compressor to pump, you can use the EQ to cut the offending frequency. Two of our voice processors give you this power: the Radius 40 has a front-panel button that lets you move the EQ ahead of the compressor, and the PS-1 allows you to repatch the order on the back panel.

Mind your s's and z's. The de-esser comes after the compressor on the 1086 and the VoiceMaster, and before the compressor on the MX60 and the PS-1. In fact, on the PS-1, the de-esser is the first stage after the input. On the MX60, the de-esser comes between the gate and the compressor.

The advantage of having the de-esser before the compressor is that you can remove sibilant peaks that cause unwanted compression. On the other hand, having the de-esser last in the chain is useful for removing increased sibilant frequencies caused by extreme compression. Unfortunately, none of the units in the group allow you to switch the order of the de-esser and compressor in the signal chain.

The de-esser on the 1086 uses a variable-frequency highpass filter to attenuate the sibilant frequencies. Consequently, it acted the most like a traditional sidechain de-esser. At the extreme settings I could hear the ducking action on sharp transients, but in normal usage this de-esser was more useful than those in the other processors. The 1086 de-esser followed the dynamic contour of the voice nicely, and at times it seemed to reach out and grab the sibilants. The 1086 also has a bar graph that visually indicates (from -1 to -30 dB) the amount of de-essing being applied to the signal.

The 1086 has a threshold and frequency control for its de-esser. The threshold is numbered from 1 to 10, and the frequency ranges from 800 Hz to 10 kHz. Having the option of going below the normal sibilant frequencies is useful for instrumental applications.

The PS-1 de-esser seemed less drastic but worked well, perhaps because of the sharp cutoff of the filter. The PS-1's de-esser uses a lowpass filter, sweepable from 800 Hz to 8 kHz. The Listen function gave this unit an advantage over the others. When Listen is active, you hear only what's above the filter, making it easier to locate the exact sibilant frequency. Boisen suggested that the precision of the PS-1 de-esser would make it harder to make a mistake with this effect.

The VoiceMaster's opto de-esser was the most subtle of the group. In fact, it's not a true de-esser at all, but rather a lowpass filter with a frequency range from 2.2 kHz to 9.2 kHz. One interesting feature of the VoiceMaster is that it includes an aux output that lets you tap the signal before it reaches the de-esser. That way you can use a de-essed signal for effects (such as a reverb), while sending the non-de-essed signal to tape. Because the VoiceMaster's de-esser is last in the signal chain, both signals will have identical amounts of dynamic and EQ effects.

The MX60 had the most easily audible de-essing effect of the group. Although Drawmer says the work is done by "intelligent circuitry," it sounds more like a shelving filter than a true de-esser, and it didn't take much attenuation to begin coloring the sound. Instead of having a variable frequency control, you choose between Male and Female tonalities. I found that the MX60's style of de-essing worked better on spoken- word material rather than on singing.

Tube or not tube. The MX60 and the VoiceMaster have features that emulate the "warm" sound of tape saturation and tubes. On the VoiceMaster, the saturation stage is before the compressor, whereas on the MX60 it's after the compressor and the EQ. A little goes a long way with each unit.

The VoiceMaster's Vocal Saturator creates smooth-sounding distortion at lower gain levels. Crank up the saturation level, and you get a delicious overdrive without a lot of noise or fuzz. The VoiceMaster gives you the option of tuning the saturation. With the Full Bandwidth button in, you get a warm, wide-band distortion. With Full Bandwidth out, you can use the Tuning control to precisely select the upper frequencies (between 1.4 and 7.2 kHz) you want emphasized. The Drive knob ranges from Clean to Unclean. I had to keep the Vocal Saturator button engaged, even when I didn't intend to use the effect. When everything but the Saturator was in the signal path, the VoiceMaster's self-noise became more noticeable. Once I added the Saturator into the signal path (with the level set to Clean), the hum went away.

The MX60 can be pushed much further than the VoiceMaster, and with the help of the 3-band Tubesound, I was able to tune in an amazing layer of distortion. It also helps that there is an extra gain control before the Tubesound stage. Having three frequency bands to work with made it much easier to get results that sounded convincing on voice. But then again, Tubesound sounded great on everything I put through it.

The most over-the-top sound came from the Radius 40 and its three tubes. With the input and compressor gains nearly maxed, and the EQ sent precompressor, I could fine-tune four frequency bands of distortion, which was enough to make the thinnest keyboard sound gigantic. You won't use this effect every day, but it's nice to know you have it.

Direct to digital. What could be handier than having digital converters built into your processor? The dbx 1086 and the LA Audio PS-1 both have optional A/D converters, though neither of the units I reviewed had the converters installed.

The dbx 1086 includes a space on the rear panel for the new 504X Digital Output card ($400). This card uses the proprietary dbx Type IV A/D converters for 16-, 20-, and 24-bit resolution. The 504X includes both AES/EBU and S/PDIF connectors, with a switch to select the format. Additional buttons allow you to select 16- or 20-bit word lengths and 44.1 or 48 kHz sample rates. The 504X also has word-clock input and output, so you can slave the unit or use it as the master clock.

The 1086's front panel has a three-position Dither switch for the converter card. In the Off position, the unit sends a 24-bit signal through the digital output. The two types of dither are TPDR and SNR2. There is also a Shape button to select Type 1 or Type 2 noise shaping.

The optional A/D converter for the PS-1, the PS-DR ($329.95), is also 24-bit capable. It has word-clock input only, but it adds a Toslink optical output to the S/PDIF and AES/EBU connectors. Like the 504X, the PS-1 can run 44.1 and 48 kHz sampling rates, but you can dither only to 16 bits. The PS-1's digital card includes an additional 11/44-inch audio input, so you can run two processors through one digital card. Keep in mind that it must be installed at the factory; you may be better off buying the PS-1D ($1,149), which has the card preinstalled.

These two units aren't the only voice processors in this price range that have digital capabilities. The MindPrint En-Voice uses the DI-Mod 24/48 card ($249), which has digital input and output capabilities via S/PDIF connectors. This means that you can use the DI-Mod to run a track from a digital recorder through the En-Voice's tube compressor and back to the digital recorder. The DI-Mod also includes an extra 11/44-inch input so that you can use both digital channels of the S/PDIF connector.

BALANCING ACT As with any piece of gear you use, a voice processor must match your style of working. Deciding on which features you need most will help you pick the processor that's right for you. If you prefer quick-and-easy results over flexibility, you'll want a unit with one or two buttons per effect. If you like to tweak, however, then features and knobs may beckon you.

HHB's Radius 40 is the easiest to use of the five voice processors. The compressor and EQ set up quickly, and the tubes give them a nice sound. The tube/solid-state hybrid mic preamp gives the voice a bit of an edge that helps it sit well in the mix. Although I didn't find the Radius 40's expander/gate that useful, and it didn't have a de-esser, I love that you can easily put the EQ before the compressor. It may be called a tube voice processor, but it works wonders on instruments, as well.

The dbx 1086 has several big things going for it: you can use the mic preamp and dynamics processor separately; the expander/gate and de-esser are fantastic; and the optional 504X A/D converter allows you to go direct to digital. The mic preamp had the most coloration of the group, but the competition in this area was stiff. The dynamics processor is easily the high point of the 1086.

The LA Audio PS-1 has the most open architecture of the five processors. The controls for each effect are very sensitive, so it may take you a little longer to get the sound that you're after. But it's worth the trouble. The EQ section is top-notch, the mic preamp sounds good, and the compressor is quite powerful once you get a handle on it. The front and back panels are easy to navigate, and I appreciate having the flexibility of using the dynamics processor, EQ, and mic preamp independently. And you can add an A/D converter to the PS-1 for direct-to-digital recording.

Drawmer's MX60 straddles the line between ease of use and tweakability. The compressor is very smooth and sets up easily; the gate is a little touchy, but it's musical once it is locked in. My favorite feature on the MX60, however, is the 3-band Tubesound, which makes this processor stand out from the pack.

The Focusrite Platinum VoiceMaster has the clearest-sounding mic preamp (it's Class A) and one of the smoothest compressors of the bunch. Each effect dials in quickly and performs well. The combination of a transparent preamp and a saturation stage gives the VoiceMaster a wide tonal palette that works well on just about anything you want to run through it.

NOT JUST FOR VOICE ANYMORE By using a single-channel voice processor rather than a mixer channel, you get the shortest distance from source to recorder and the best signal-to-noise ratio possible. And because a voice processor can be just as useful for mixing as for recording, it's a worthwhile investment.

Gino Robair is an associate editor at EM. Special thanks to Myles Boisen, Jill Garellick, Jeff Casey, Steve O., Brian Knave, and Laura Forlin, and to Elliot Garellick for his patience.