Remix is a plug-in that "uses cutting edge technology based on artificial intelligence" in order to un-mix a stereo track into five stems of Vocals, Piano, Bass, Drums, and Other. There have been a handful of apps that do this recently, but Remix is the first I've seen that can be inserted on a source track and adjusted in real time. The ability to set the Sensitivity levels, solo stems, and adjust stem Gain levels while auditioning is a real plus, as it's something I'd previously had to guess at and run multiple passes of separation in order to get the results I desired with other software. As in the case of all track separation apps, when soloed there are usually a certain amount of audible artifacts, but when recombined the mix will sound correct. The controls are fairly simple. Along with the aforementioned Sensitivity and Gain, there are Solo and Mute buttons on each stem, plus controls for storing presets. Controls and such can be automated if needed, which could be handy. Think of the Sensitivity controls as gates for the desired stem element. If turned up (clockwise), it'll sound more natural and bring more of the element in, but it will also have extra spill from other sources. This is where the real time ability comes in handy, and picking the exact settings really does change the outcome.
One of my students, Jeff Frankel, had just sent me a stereo recording from a series of live club shows his friend had videoed, and he wondered if I could teach him how to fix the balance, so I put Acon to the test. By adjusting the Vocal Sensitivity to lower settings (counterclockwise) I could dial out some of the crowd's talking, which was a relief! Drum tracks retained far better low end and kick drum than other programs I have tried, and that really helped as well. I'm not sure why "Piano" is one of the main stems that gets separated. In this non-piano music, it picked up some electric guitar parts but not all, so combining Piano and Other stems was where the chordal information seemed to be. I could get a better mix, but still the reverberance of the rooms these songs were tracked in, and the incessant crowd yammering, were issues that I could only overcome to a certain level. But Remix did give me enough control to build a better image of the band in the room, and clearer vocal parts.
Because of the operations going on, as I expected there's a fair amount of "acceptable latency," as Acon puts it. This shouldn't matter unless you're trying to overdub to the separated tracks, but as a test I simply duplicated five instances of my source track and rendered all the stems with the desired track in Solo on the plug-in. This rendering is far faster than other separation software I've used. Some DAW plug-in formats (AAX and VST3) will allow you to output each of the five stems for processing on aux tracks being fed from the original file. This worked well in Pro Tools, although if I hadn't looked this feature up in the manual I would have had no idea it existed.
I'm always excited to try new tools that help with audio restoration. When I think back to the number of stereo live concert cassette board tapes out there with overly loud vocals and direct input instruments, and how software like this can rebalance them, I get excited. Remix is the kind of technology we wished for decades ago, and it's very affordable. Now, let's get busy fixing some old (and new) audio!
_disp_horizontal_bw.jpg)