Audionamix’s ADX Vocal Volume Control plug-in (VVC) addresses a common, critical, and chronically unmet need of producers and mastering engineers: adjusting the lead vocal’s volume in a baked-in stereo or mono mix of a multitrack production without affecting other elements in the mix.
Sure, you can boost the lead vocal in a stereo mix using mid-side processing to raise the volume of the mid channel—but not without also boosting the kick, snare, bass guitar, and any other center-panned elements. (And mid-side processing is worthless on mono mixes and stems.) VVC promises more discrete—arguably magical—processing. If it works, I thought to myself, this would be the Holy Grail and the answer to my prayers. I was dying to give it a spin.
Version 3.0 of the cross-platform plug-in is available in AAX Native (64-and 32-bit), AU, and VST formats. I reviewed the AU plug-in in Digital Performer 9.02 (DP), using an 8-core Mac Pro running OS X 10.11.5.
HOW IT WORKS
VVC uses multiple algorithms to separate lead vocals—or any monophonic melodic instrument—from a track, whether the original file was stereo or mono. After the separation is complete, you can change the vocal or instrument’s volume and pan position ostensibly without affecting other elements in the mix.
After instantiating VVC on a track insert, you make a selection of the audio you wish to be separated. (In DP 9.02, selecting the audio is unnecessary.) Then you either perform an offline bounce or play through the selection so the plug-in can acquire the audio. Pressing the Separate button sends the data to Audionamix’s ADX servers, where it is processed. Access to the ADX cloud is granted by an API (application programming interface) key you receive after purchasing the software (iLok is not used).
Fig. 1. Audionamix ADX VVC 3.0’s simple interface makes it easy to adjust levels and panning of vocals and monophonic instruments in a bakedin stereo or mono mix. Use the Separation Mode button on the left side of the GUI to select vocal or instrument processing. The plug-in’s Separation Mode determines what material gets separated from your audio selection (see Fig. 1). In Vocal mode, the proprietary processing includes Audionamix’s Automatic Voice Activity Detection (AVAD) algorithm. AVAD detects where singing occurs in the selected material and extracts melodic content only in those sections, leaving the original mix unaffected elsewhere. In Melody mode, melodic content is extracted throughout the entire audio selection; this mode is not only useful for extracting monophonic guitars and the like, but also as fallback processing should AVAD fail to extract some vocal phrases. In either mode, activating the High Quality button will yield better-sounding results (at the expense of a longer wait for processing to complete).
When separating vocals, enter the highest and lowest notes that were sung in the GUI’s two Pitch Range boxes to help VVC target the vocal more precisely. If you want to separate reverb in your original file along with the vocal, turn on VVC’s Reverb option before engaging separation processing. Doing so will maintain the wet/dry balance of the original vocal as you make subsequent changes to its gain and pan position. Activating the plug-in’s Consonants option helps VVC to extract the high-frequency, noisy components of a vocal produced by enunciated consonants; if you hear drum hits or other non-vocal transients being separated with the vocal, turn this option off and engage the processing again.
After the vocal has been separated, you can use VVC’s controls to boost or attenuate it up to 12 dB and pan it up to 60 percent of the way toward a hard-panned left or right position in the stereo field. Volume and panning adjustments, as well as the plug-in’s bypass, can be automated; for example, you might want to automate the volume to correct an overly dynamic vocal.
A TRIAL SEPARATION
My acid test of VVC 3.0 was to boost lead vocals on a 2-track, single-song master recorded in 1981, featuring the as yet undiscovered Ashley Cleveland singing. Ashley’s stunning vocal had been mixed too low in the stereo mix, the only surviving recording of the sessions. I was fervently hoping VVC could boost Ashley’s riveting performance.
Even in High Quality mode, VVC 3.0 demanded virtually no local CPU resources in DP—no doubt due to all separation processing being carried out in the cloud. The disadvantage of cloud-based processing is that, if your Internet service goes down during a session, you’re SOL.
Fig. 2. Another view of Audionamix's interface, this time with new pitch range selected, the Reverb Separation Options buttons de-activated, and the Separation mode set to Melody. Vocal separation, for the two minutes of the track that contained vocals, took around ten minutes to complete in High Quality mode. Afterwards, I boosted the lead vocal 6.5 dB in VVC’s GUI. Unfortunately, doing so also boosted center-panned lead guitar fills and its effects. However, the electric bass, drums, and Mellotron—all of which were present in the center channel—were not boosted. The Mellotron, playing a string pad, had been masking Ashley’s lead vocal a bit on the original stereo mix, so being able to boost her vocal above the level of the pad was a major coup (not to mention astonishing). And activating VVC’s Reverb button before separation preserved the wet/dry balance of analog plate (hardware) reverb on Ashley’s boosted vocal.
The only artifact I heard on the boosted vocal and guitar was an occasional, fleeting boost in the upper-bass band, which would be easily addressed later using third-party multiband compression. Most important, I heard no phasiness—the boosted vocal sounded completely coherent and intact. I was blown away.
After separation, some vocal lines remained quieter than others. Automating VVC’s volume slider to correct this, the lack of a numeric readout while adjusting the slider made precise real-time adjustments difficult. (A readout appeared while hovering my mouse over the stationary slider but not while adjusting it.) Also, automating the volume slider sometimes arbitrarily activated the plug-in’s bypass; oddly, I could fix that issue simply by rewinding DP and playing through the affected section again. Most important, I could make transparent real-time adjustments in the lead vocal’s level over a 3dB range (boosting between 6.5 and 9.5 dB); it was like using a time machine to go back to the original tracking session and ride the singer’s fader. Amazing!
TAKING THE LEAD
After getting Ashley’s lead vocals dialed in, I turned my attention to the lead guitar, which was mixed too low in places. To preserve VVC’s vocal processing, I made a copy of the original track and instantiated VVC on it in Melody mode to process the guitar. As the guitar sounded a bit too wet with effects in the sections I wanted to boost, I deactivated VVC’s Reverb button; that allowed me to boost only the guitar’s dry signal, making the guitar sound clearer. I could also pan the guitar in the stereo field using VVC’s pan control. I was happy to note that volume automation I recorded on the track survived implementing the separation process a second time using different plug-in settings.
Once I had the lead guitar’s levels automated, I comped the two copies of the track—containing respective VVC-processed vocal and VVC-processed guitar—in sections to produce a single track with automated vocals and guitar. The end result was a huge—and miraculous—improvement over the original mix.
Make sure you bounce your tracks after applying VVC processing; in DP 9.02 running El Capitan, pre-existing real-time VVC processing on a track did not endure after a reboot, and tweaking the plug-in’s volume and panning controls had no effect until I reran the separation process. VVC 3.0 is a bit buggy, but that shouldn’t stop you from buying or renting this groundbreaking (and inexpensive) plug-in. If you’ve got old masters gathering dust because they were poorly mixed and not up to today’s commercial standards, ADX VVC 3.0 just might be the miracle cure you need. Every engineer who does mastering, remastering, or restoration work should own this pioneering plug-in. (As we were going to print, Audionamix told us about their speech-enhanced plug-in, ADX SVC, scheduled for release later this year. It is being designed to provide independent volume level control over both speech and background elements within a mono or stereo mix.)
Sounds remarkably transparent. Can preserve the wet/dry mix of adjusted elements. Automatable. Easy to use. Negligible CPU loading. Inexpensive.
Processing has trouble discerning between lead vocals and center-panned electric guitar. Volume slider lacks numeric readout during adjustment. Real-time processing is volatile. Cloud-based processing not available during Internet outages. A bit buggy.
Michael Cooper is a recording, mix, mastering, and post-production engineer and a contributing editor for Mix magazine. You can reach Michael at firstname.lastname@example.org and hear some of his mixes at soundcloud.com/michael-cooper-recording.