Issue link: https://digital.copcomm.com/i/933783
C A S Q U A R T E R L Y W I N T E R 2 0 1 8 35 IN THE STUDIO: AUDIONAMIX'S TRAX PRO 3 AND XTRAX STEMS I had the opportunity to check out the most recent version of the Audionamix software "TRAX Pro 3" and its newest colleague "XTRAX Stems." Both utilize a form of machine learning called "neural networks" (described in depth in this issue in "Neutral Networks: A New Way to Think About Processing" by Jay Rose CAS). While it's a given that every show has its own set of challenges and opportunities, one of the recurring issues I wrestle with is integrating music: How can I better control the music so that it can help support the emotion of the scene without becoming distracting? Getting mix stems is, of course, the first and best place to start. But stems may not be available, and stems sometimes bring as many problems with them as they solve. Commercially released songs often aren't available with splits and, on those occasions when stems are available, they can lack the luster found in the original master. For better or worse, we often find ourselves attempting to gracefully weave the two-track master of a song into a story without the direct control of elements we might desire. Audionamix has an interesting and, perhaps, unique approach that can help tackle this dilemma. The company's TRAX Pro 3 and XTRAX Stems utilize remote servers by Stephen Fitzmaurice CAS loaded with neural network algorithms to analyze and separate the melodic content of a song or cue—or even dialogue—from the surrounding material. In general, TRAX Pro 3 can separate the program into speech (e.g., vocal, dialogue) and "the rest," while XTRAX Stems separate the file into drums, voice, and "music" (i.e., everything else). The idea being that you can then have greater control over these groups. For re-recording mixers, if we're given a music track with vocals that has to play under dialogue, we have to find a way so the two don't fight for frequencies. Eq-ing the vocal range of the music track when dialogue is present is a common approach. With this software, the ability to dip the lead vocal down under dialogue more independently is an attractive and worthwhile feature. However, can you really unbake a cake? Well, no. Push too hard and you'll definitely end up with distracting artifacts. I'll admit, though, I was pleasantly surprised by my newfound ability to emphasize or reduce elements within a mixed track, and TRAX Pro 3 gives the user multiple levels of editing ability to increase the accuracy of the separation even further. I'm always wary of relying on manufacturers' demos, as they are often strategically chosen. If you're curious, however, their website Audionamix. com includes a number of clips that show off the software's capabilities, which are impressive—even if they are optimal examples. YouTube has a handful of videos demonstrating additional examples by "real" users and, as mentioned in the article, Jay has posted some examples as well. I'll continue digging deeper into the features, strengths, and weaknesses. But for now, I'm happy to report that I am regularly able to enjoy significant improvements in sound separation not always achievable with other technologies. • Audionamix TRAX Pro 3 Audionamix XTRAX Stems