Antares have updated their Avox vocal processing bundle and, alongside further tools for character manipulation, automated harmony generation is now included.
However good your arrangement, instrumental performances and mix might be, without the right vocal even the best song can struggle to survive. Of course, what constitutes the 'right' vocal is a subjective matter, but the ability to creatively adjust the raw vocal performance is something we almost take for granted in our modern DAW–based recording systems. Antares have been at the forefront of vocal manipulation, and their original Avox bundle (reviewed in the January 2006 issue of SOS: www.soundonsound.com/sos/jan06/articles/antaresavox.htm) provided a collection of five plug–ins. Choir and Duo simulated doubling and multitracking, Sybil was a straightforward de–esser and Punch a simple compressor optimised for vocals, while the Throat plug–in was an ambitious attempt to model the vocal tract, allowing the user to change various characteristics of the virtual throat and, therefore, the character of the voice. While no one was claiming Throat could turn a Britney Spears into a Robert Plant, for subtle shifts in vocal character and gender it could work very well, with extreme settings providing fun 'special effects'.
Providing your expectations of what was possible with Throat were realistic, perhaps the only down side of the original Avox bundle was the cost: some may have baulked at an asking price of £400. Now Avox is back, and not only is the UK price somewhat lower, at £350, but Antares have included five new plug–ins alongside those of the original bundle. So what extra vocal processing options does the new Avox 2 package provide, and do they add significantly to the value offered?
The new Harmony Engine builds on the principles of the Choir and Duo plug–ins to create a processor that can generate full four–part harmonies, in much the same way as hardware and software products from Digitech, TC–Helicon and Mu Technologies can. Straightforward creative processing is provided by Warm — a belated replacement for the now defunct Antares Tube valve simulator — and vocoder–like Articulator, while Mutator and Aspire concentrate upon further voice modelling processes. The five plug–ins that comprised the original release appear little changed and, given that these were described in some detail in Paul White's earlier review, I'll concentrate here upon the new additions.
The plug–ins are for Mac and PC and are supplied in RTAS, VST and Audio Units.Avox 2 is available as a boxed version or a downloadable product directly from the Antares web site, in both cases requiring an iLok key (not supplied) for authorisation. Both installation and authorisation proceeded without any difficulty on my test PC, and I was up and running with the VST versions of the plug–ins inside Cubase 4 within just a few minutes.
Of all the new plug–ins, Warm is perhaps the most straightforward. The user interface follows the 'house' style and, while the controls are simple, there is still plenty of flexibility. Two tube styles are offered: Velvet does smooth preamp impersonations, whereas Crunch is more of a valve guitar amp simulation. Unless Omnitube is engaged, the tube effect is only added when the level enters the upper two segments of the input meter, but the Gain slider makes it easy to control how much this happens. The Drive slider does exactly what you might expect.
To my ears, at least, Warm is one of the nicest tube emulations around. The Velvet model provides an extremely smooth warming effect on vocals and the interaction between the Input and Drive levels allows this to be super–subtle or up front and obvious. Either way, the sound quickly becomes quite addictive, and some A/B listening is required to ensure that it doesn't get overcooked. If overcooked is what you want, whether on vocals or instruments, Warm can also do that with a combination of Crunch, Drive and the Omnitube switch — great for industrial or lo–fi sounds. The plug–in makes modest CPU demands and, as it supports both mono and stereo inputs, it can also be placed across a full mix, although I found that this required considerable restraint if the results were to remain transparent.
Articulator is the Antares take on a vocoder and, like all such devices, requires two audio sources. In Antares' terminology, the modulator (usually the vocal) is called the Control signal, while the carrier is known as the Audio signal; this is the target audio to which the Control signal's formant and amplitude properties will be applied. The plug–in includes a built–in noise generator that can be used as the Audio signal, but if you want to use external audio as a carried source, Articulator requires a DAW that supports side–chaining. This is fine for Pro Tools (RTAS) and some AU–based hosts, but Avox 2 was developed before Steinberg introduced more flexible audio routing in VST3, so a workaround is required when using VST–based hosts. This involves Articulator being fed from a stereo track where the Control signal is in the left channel and the Audio signal in the right. The process is a bit clunky, as you have to create such a track prior to applying the plug–in — something that might need repeating a few times until you find the right combination of Control and Audio material — but it does work. This aside, however, Articulator is capable of delivering a good range of vocoder–style processing, and the various formant, amplitude and EQ controls provide plenty of ways to customise the interaction between Control and Audio signals. Hopefully, Antares will update Articulator to the VST3 format, as this would certainly make it much easier to use for Cubase users.
Mutator's voice–mangling process is based upon ring modulation: pitch detection is used to match the pitch of the ring modulation to that of the original signal, and the Mutation control adjusts the relative frequency of the modulation. The Alienize process splits the audio into short sections and then reverses these during playback. This processing becomes more extreme at higher Dialect settings, gradually decreasing the intelligibility of the vocal until it really does sound like some bizarre alien language. When the Voice Design controls are added into the mix, it is clear that Mutator can go from subtle voice changing through to sci–fi special effects. Whether you need a Darth Vader effect (pitch-shift and Mutation do a decent job here), space aliens (everything on full), or just a subtle alteration of a spoken voice to protect the identity of an interviewee, Mutator produces some useful (and downright entertaining) results. Yet when the Voice Design section is used in isolation, the plug–in is also perfectly capable of basic pitch– and gender–changing of sung vocals, and is easier to use than Throat.
The Aspire plug–in attempts to modify the 'breathiness' or aspiration noise of a voice. The basic idea is not dissimilar to the Breath and Growl options provided in TC–Helicon's more recent voice modelling products, although I think the range of processing offered by Aspire is a little less dramatic. Aspire works by separating out the aspiration noise from the harmonic element and then allows the user to change the amount and tonal character of that noise. This does, of course, depend upon there being some aspiration noise in the first place. A singer with a very pure, clean voice will generate minimal aspiration noise (in comparison with a Joe Cocker impersonator) and the plug–in will have little to work with.
Interestingly, the plug–in can be used to both add and subtract breathiness and, in use with a range of vocal types, I was pleasantly surprised at how well it worked in both modes. For example, it did a very effective corrective job of cleaning up a female vocal part that had just the slightest hint of breathiness, but it could also be used with the same vocal to add a bit of extra rasp. This was a subtle change in either direction and, unless the Increase control was maxed out, the results were very natural. The ability to target any aspiration increase to a particular frequency band also seems to help in this regard. Used with a more whisky–and–cigarette–soaked blues vocal, the results were equally good. A surprising amount of cleaning up could be achieved, while the voice could also be pushed a little harder, although with the greater aspiration noise present in the original vocal it was easier to overdo this and move into 'broken microphone' territory. The bottom line here is that Aspire actually works quite well, provided you use it for subtle changes. Like all the Avox plug–ins, Aspire's controls can be automated, and varying the degree of breathiness can be used to add some nice character variation through a performance.
Which leaves us with Harmony Engine and, without giving the game away too much, I've saved the best until last. In a nutshell, Harmony Engine provides up to four automatically generated harmony voices, each of which can have various humanising and voice–modelling options applied to give the backing vocals a more natural feel. The harmonies themselves can be triggered in a number of different ways, including fixed or scale intervals, chord forms (including inversions), automatic chord recognition via MIDI, or MIDI notes. In terms of automated harmony generation, this is probably as flexible as it gets at present.
In use, Harmony Engine requires an input signal that is as clean as possible — a monophonic voice with a reasonable level and free from background noise. De–essing, compression and good pitching (via pitch correction, if required) are helpful. The rather intriguingly labelled 'Trial & Error' slider on the input channel controls how Harmony Engine tracks pitch and, regardless of how well my input signal met the above requirements, I always found it useful to experiment with this slider to locate the optimum setting for a particular vocal performance. If the lead vocal was to be used as the input signal (as opposed to a separate vocal specifically recorded to drive the harmonies), I also found it easier to create a duplicate track specifically for use with Harmony Engine. This allowed me to process the signal specifically for best results with the plug–in, while the original might be processed in a different fashion to suit the overall mix. The Mute button on the Input channel allows the input signal to drive the harmony production without being heard at the Harmony Engine output.
The operation of some of the harmony modes is fairly self–evident but, like Mu Technologies' Mu Voice, Harmony Engine provides the user with excellent flexibility via the use of Harmony Presets and Voice Parameter Presets. The Voice Parameter Presets system provides six slots that can store the current state of the Harmony Channels and Humanize/Glide controls. Switching between these presets can be done manually or via automation and provides a rapid way to move between, for example, a simple voice double and a four–part harmony. The Harmony Presets come into their own in the various chord modes, where each of the 15 slots can be used to define a chord type. Chord Name mode allows specific chords to be defined, as shown in the screen shot, but there is also a Chord Degrees mode in which the presets are all configured relative to the key of the song, meaning that the presets automatically adjust themselves if you change the key. The other clever element of these presets is the Spread and Register controls, which change the way in which Harmony Engine generates the harmony parts for a particular chord. Spread defines the pitch spread over which the harmonies can be spaced, while Register sets the general range of the lowest harmony note. These settings are stored with each preset and make it very easy to create the required combination of wide or narrow harmonies and high– or low–pitched harmonies. If the same chord is assigned to two slots with different Spread and Register settings, switching between them (manually or via automation) can add some really effective variety to the generated harmonies.
The other interesting feature is the Freeze function, which is similar to Harmony Hold in some of the newer TC–Helicon harmony products. It provides two different modes: with Formant Only, the current vocal sound is sustained but pitch continues to vary, while in Formant + Pitch mode, both vocal sound and pitch are held until the button is deactivated. These modes can be switched on and off manually or via automation, but the timing is critical to make sure the sustained vocal sound falls naturally on a vowel. Both modes have their uses and provide a nice way to break the direct link between the input vocal and the harmony.
Getting useable results out of Harmony Engine does require a little practice but, in this respect, it is no different from any of the competing auto–harmony products. In part, the issue is one of arrangement — what harmonies are required and where — and even with the intelligence built into the engine, a little bit of musical commonsense is required. The second issue is simply giving Harmony Engine the best opportunity to do its stuff by supplying it with a high–quality input signal. Like all the competition, it is easy enough to produce glitchy output if the input signal is noisy or poorly pitched, but with the right input signal the audio quality of the harmony parts generated by Harmony Engine is as good as it gets at present. For simple double– or triple–tracking, Harmony Engine is both easy to use and very convincing, and for more complex harmonies, at their best, the results can still sound very good even when taking a prominent position in a mix. Placed in a background context, I suspect most average listeners would not even register that these are auto–harmonies rather than a group of real singers. And as it's a studio–based tool, you can, of course, keep tweaking to optimise the quality of the output and reduce any processing artifacts to a minimum.
Whatever your take on the original Avox bundle, there is no doubt that Antares have really upped the ante with Avox 2. The five new plug–ins make this a significant upgrade and offer at least as much as the original five. For me, Harmony Engine is the real highlight. With a little perseverance, it is capable of extremely realistic results and, for a home/project studio owner without the resources to book a group of session singers on a regular basis, it provides a viable and practical alternative. Of the other plug–ins, what Aspire attempts to do is also very interesting, although its effect has to be used in a subtle fashion, while Warm is one of the best tube emulations I've used for vocal processing.
All of the plug–ins within the Avox 2 bundle are available individually and, clearly, whether the package represents a good buy will depend upon which of the processing options you might be interested in. That said, as a 'vocal processing toolkit', and sat alongside your choice of pitch–correction software, Avox 2 does provide a lot of creative and corrective possibilities. By including the impressive Harmony Engine and pitching Avox 2 at a slightly lower asking price than the original Avox bundle, Antares have significantly enhanced its appeal. The options provided by Avox 2 are powerful, intriguing and fun in equal measure. If you find yourself struggling for vocal processing options, you really should check out the audio examples and time–limited demos on the Antares web site.
- Mac: 1.5GHz G4 processor (1.8GHz G5 or Intel processor recommended), Mac OS 10.4 or later, suitable RTAS, VST or AU host, iLok key.
- PC: 1.7GHz single–core processor (dual–core processor recommended), Windows XP or Vista (as required by host), suitable RTAS or VST host, iLok key.
If you are interested in both voice modelling and harmony generation, perhaps the only direct comparison is with products from TC–Helicon. For example, the hardware VoiceWorks Plus offers a combination of pitch correction, harmony generation, voice modelling and effects such as dynamics and reverb, and includes a mic preamp. It is, however, more expensive than the Avox 2 bundle, although TC–Helicon also offer a number of less expensive alternatives for specific tasks, and users of TC's Powercore DSP products can buy into software equivalents of the same processing. For harmony generation, the other obvious competitors are the hardware–based Digitech Vocalist range (aimed at both live and studio use) and the software–only Mu Voice from Mu Technologies (reviewed in the July 2008 issue). Mu Voice is a relatively new product from a relatively new company but its feature set is not a million miles away from that of Harmony Engine, and it would be worth auditioning alongside the Antares product if automated backing vocals are your primary need.
- Harmony Engine is about as good as it currently gets for automated backing vocal generation.
- With twice as many plug–ins as the original version, Avox 2 offers very good value for money.
- Not all users will want or need all the plug–ins.
- Articulator does not, as yet, fully support side–chaining under VST.
Avox 2 provides a bigger bang for the buck than the original bundle, and Harmony Engine represents a significant addition. A very attractive collection of plug–ins providing both corrective and creative vocal–processing options.
information£349.99 including VAT; upgrade from Avox $199.
Sonic8 +44 (0)8701 657456.
- Antares Avox 2 v.1.0
- PC with Athlon dual–core 4400+ and 4GB RAM, with TC Electronic Konnekt 24D interface, running Windows XP Pro SP2.
- Tested with Steinberg Cubase 4.1.3.