The makers of VocAlign have built on its capabilities, creating software that can not only match the timing of vocal parts, but their pitch and level too.
Regular readers of SOS will be familiar with the Synchro Arts name through their VocAlign software, which was originally reviewed back in July 2000 (/sos/jul00/articles/synchro.htm). VocAlign was designed to help engineers performing Automated Dialogue Replacement in film or TV contexts, where on-set live sound needs to be replaced with later studio voice recordings. As the name suggests, VocAlign's primary function was to extract the timing from the original voice track and manipulate the replacement track so that the new audio lip-sync'ed as tightly as possible with the original and hence the visuals. Of course, music producers have also found plenty of uses for VocAlign in tightening up multitracked vocal parts and bringing doubled-tracked leads or multiple harmony vocals tightly into time.
Synchro Arts have recently revamped, expanded and rebranded this technology, creating a new product called Revoice Pro. While this new product retains VocAlign's ability to match the timing of one performance to that of another, RVP2 also promises to match the "energy” (level) and pitch: quite an ambitious feat.
RVP2 is available as a download from the Synchro Arts web site, and authorisation uses the iLok system. A limited-time trial licence is available on request if you want to try before you buy. The Synchro Arts web site also provides a selection of useful audio examples and access to a helpful web-based user guide, although, unlike a PDF, this does mean you need to be online to consult it. RVP2 is a stand-alone, Mac-only application, but integration with your DAW is made easy by AudioSuite, AU and VST3 'link' plug-ins that allow you to transfer audio from a track within your DAW into a track within RVP2. Once processed, audio can be dragged and dropped back into your host DAW. The software supports sample rates between 44.1 and 192 kHz and word lengths of 16 to 32 bits. The only constraint is that all audio files open in a single RVP2 session have to be of an identical audio format.
Although RVP2 includes the ability to generate a brand-new double track from a single vocal performance, operations tend to start with a minimum of two original recordings: the 'guide' and 'dub' tracks. In an ADR context, the guide is the original on-set dialogue that you aim to replace, while in a musical context, the guide represents the best of your vocal takes; this won't be replaced, but your double-tracked or backing vocal tracks need to follow it. The dub track is, respectively, either the replacement dialogue that needs to be synchronised, or a second vocal take that you want to sync a little more tightly to your best take. For musical applications, you might, of course, have several 'dub' tracks such as multiple backing vocal takes, for example.
RVP2 provides a track-based working environment. Users can add new tracks as required to hold the guide, dub and any output audio from RVP2's processing. The user interface is generally straightforward in use, but does have a few quirks; for example, there's no Play button alongside the other transport-related controls such as Loop and Rewind, and instead you use the space bar to toggle playback on and off. Using the 'link' plug-in to import your guide and dub audio into RVP2 is easy, and the VST3 version worked very smoothly with Cubase 7 in my own testing. The plug-in is placed as an insert effect on any track containing audio you wish to transfer to RVP2. You then simply specify which RVP2 track the audio should be passed to, engage the Link button in both the plug-in and the main application, and then set your DAW to play back the timeline section you wish to transfer. If required, the AU and VST3 plug-in versions allow you to transfer multiple tracks from your DAW in a single pass, which is very useful if you have multitracked vocal parts that you want to work on within a single RVP2 project.
Once your audio is in RVP2, in order to process it you first have to select a portion of the guide track to work with, and then define a 'process'. Two basic types of process are available: Doubler and Audio Performance Transfer (APT). A Doubler process creates an automatic double-track from just the guide track (no dub track is required), and you can choose either a mono or stereo output. Via the Process settings dialogue, you can also tweak how tightly the timing matches the original guide performance, shift the formants to give the double a somewhat different character and adjust the amount of vibrato. Once you've mastered the mechanics, it takes just seconds to execute the process with the Render button and audition the results, and if you want to tweak further, you just adjust the settings and re-render. This is about as painless as automatic double-tracking could possibly get.
The APT process requires both a guide and a dub track. The processing technology going on here is pretty clever: the timing, pitch and energy or volume of the guide track are analysed and then applied to the audio on the dub track. Again, the user can define how much of each of these elements they attempt to transfer to the dub. So if you want to apply just the timing, as VocAlign does, you can disable the pitch and level transfer entirely. Equally, if you just want to apply a proportion of any of these elements, the Transfer Strength setting for each of them can be adjusted. In principle at least, this allows you to control exactly how far the dub track is processed to get it to match the timing, pitch and level of the guide track.
Unsurprisingly, if you give RVP2, as a starting point, guide and dub tracks that are vastly different in timing, pitch or level, and expect it to produce artifact-free results, you will be disappointed. However, if you start with a guide and dub that are even moderately close, RVP2 will generally get things as tight as you want (or not so tight, if that's a better creative option) without breaking sweat. Again, if you don't like the results, you simply tweak the settings further and re-render; quick and entirely painless.
With more problematic audio, RVP2 allows you to place multiple 'processes' into a project and work on sub-sections of your audio separately. There are also some basic waveform editing tools that allow you to adjust the audio — perhaps manually shifting a word or phrase or cutting out non-essential noises such as breaths — to make the software engine's task a little easier. That said, I found it more efficient to do this kind of editing with my familiar DAW toolset, including any pitch correction on my guide vocal, prior to transferring the audio into RVP2.
Moving processed audio back into your DAW is very straightforward, involving a couple of key presses and a drag and drop. In Cubase 7, this worked a treat, and, providing you're working with WAV files (as opposed to AIFF, which doesn't contain timecode stamps), the Cubase 'Move To Origin' command will automatically position the dropped audio to exactly match the time location of the audio used in the guide track.
In testing RVP2 with vocals, as well as the obvious applications, I also tried pushing things a little further by matching a male dub to a female guide and getting a sung vocal dub to follow a guitar melody used as the guide. For the more conventional tasks, the results were truly excellent. The Doubler process is perhaps the most efficient and effective way I've ever encountered to generate automatic double-tracks. The stereo option is particularly effective and, for example, can instantly give a lead vocal the extra body required to lift a typical pop chorus. The only danger here is that the results are somewhat addictive! With the APT process, the only additional qualification is the obvious one; in terms of audio quality, good results are easier to obtain if the phrasing of the guide and dub are a reasonable match to start with. RVP2's processing is then able to get you exactly the tightness (or otherwise) that you might need, without any obvious audio artifacts.
As might be expected, the results of my more experimental tasks were somewhat less predictable. However, the 'female to male' APT, while needing a little fine-tuning to develop the most appropriate settings, worked remarkably well. If you need to tighten a mixed-gender set of backing vocals, providing they have been sung reasonably well in the first place, RVP2 can make the task easy. Trying to get a vocal to track an instrument proved, perhaps not surprisingly, more challenging, but I was still surprised at just how far I could go, and it was interesting to hear the voice following the pitch-bends in the guitar part. While you shouldn't perhaps expect RVP2 to fix a dodgy singer using a recording of a skilled guitar player, more creative souls could have some fun here, with possibilities that include some interesting vocoder-style effects.
When it comes to dialogue replacement, RVP2's APT process is remarkable. Whereas VocAlign could generate excellent lip sync, RVP2 goes much further, by allowing the user to apply as much (or as little) as required of both the energy and pitch of the original guide vocal to the dub. The end results can be utterly convincing, unless the dub track contains very different intonation to the original, in which case you can push things too far and expose the processing. In those cases, you just have to accept that some of the pitch and level settings have to be dialled in more subtly.
While much ADR work involves the original actor re-recording his/her lines (and RVP2 will generally deal with this without breaking sweat), what's even more remarkable is how well the software will match the voice of a different 'dub' actor to the guide. For film and TV audio engineers who work with dubbing foreign language versions of films, RVP2 could become something of a silver bullet.
RVP2 is obviously a niche application: its raison d'être is to match the timing, pitch and level of one audio signal to that of another, and — qualifications aside about how hard you can push the processing if the guide and dub signals are very different to start with — excellent results can be achieved very efficiently. However, do note that while RVP2 can match pitch and timing between two audio tracks, it doesn't correct pitch in the same way as Auto-Tune or Melodyne. What RVP2 does is perhaps best described as 'pitch matching' rather than 'pitch correction'.
At £499$799, this is unlikely to be a casual purchase for most people (although at the time of writing, the introductory price is 25 percent less than this and there are also upgrade paths from VocAlign), so who should be putting RVP2 on their shopping list? Obviously, anyone doing dialogue replacement professionally on a regular basis will want to investigate Revoice Pro sooner rather than later. It is also a very interesting and powerful tool in a musical context for generating doubled and multitracked vocals, and getting backing vocals to fit with a 'master' track. Whether it's a 'must have' for this function is a more difficult call, as most DAWs now have something in their armoury that allows you to attempt this kind of thing; some, like VariAudio in Cubase 7, even allow you to re-pitch the double-tracks to create harmony parts. RVP2 doesn't allow harmony generation, but for automatically generating and tightening up double and multitracked vocals, the results are as good as I've heard from software. If you work with vocals a lot, I suspect the same efficiency argument applies as for ADR work. Whether it is for ADR or vocal production, if your budget still has a little slack in it, downloading RVP2 and requesting a trial licence will soon allow you to evaluate the cost-benefit relationship for yourself. Revoice Pro 2 may well be a niche tool, but within that niche, there is really nothing to match it.
- Could bring considerable efficiencies for ADR or vocal production professionals.
- Capable of producing excellent results with both dialogue and vocals.
- Price means it is unlikely to be a casual purchase for most.
- Slightly quirky user interface, though still efficient to use.
For ADR and vocal production professionals, Revoice Pro 2 promises some considerable workflow benefits, and the results it achieves can be truly excellent.
- Apple iMac with 3.4GHz i7 CPU and 16GB RAM, running Mac OS 10.7.5, with TC Electronic Konnekt 24D interface.
- Tested with Steinberg Cubase 7.0.1.