By Michael Raine
This article originally appeared in the July/August 2015 issue of Canadian Musician
Depending on who you listen to, drag-and-drop, cloud-based mastering is either the great equalizer of the recording industry, or it’s a con job that further degrades the quality of music. Around the globe, from Los Angeles-based (and Justin Timberlake owned) AfterMaster Labs to Spanish company MasteringBOX, services are popping up that promise to make the “black art” of mastering unfathomably simple, quick, and affordable. But these companies are all playing catch-up to one Montreal-based start-up, MixGenius, and its flagship product, LANDR.
When people speak about tablets, they’re primarily referring to the iPad, and similarly, when there is discussion of online instant mastering, the talk focuses on LANDR. And there has been a lot of talk.
Fast Company magazine included it on its recent list of “the music industry’s next big disrupters” and Vice’s Noisey says “LANDR might be the biggest leap forward in DIY recording technology since the development of home multitrack recorders.” The company’s website features quotes of glowing praise from the like of Tiga, of Montreal’s Bennett Lewis, and Sie Medway-Smith, engineer for U2 and Bjork. On the other side, it’s not difficult to find articles, forms, and videos of musicians and audio pros criticizing LANDR.
So who’s right? Can an algorithm really take a track you’ve recorded in your project studio and achieve “results that rival professional studio work in minutes?” Or is it a flawed concept that will never live up to its promises? You’ll have to come to your own conclusions, but to get you thinking about it, we’ve sought out one of the driving forces behind LANDR, an end user, and one of its biggest critics.
The Creator**Justin Evans, VP of Products & Innovation at MixGenius**
The origins of LANDR begin when Justin Evans, with the backing of some seed money, visited the Digital Music Centre at the Queen Mary University of London in 2012, where a group of PhD students and researchers had been developing this new technology. Evans compares the technology behind LANDR to that behind the “listening” app, Shazam. “What Shazam does when you hold it up is it’s actually breaking this music into features and then comparing that against a huge, huge database,” says Evans. “So when you put a track [into LANDR], it analyzes what kind of music it is, what the main musical features are, how that makes kind of a finger print against it, and what the appropriate processes are to apply, and then some combination of that and musical analysis that understands the frequency content that’s much less about the type of music and more about what’s in your mix. Those two systems speak to each other and then, basically, do the mastering techniques for it using all the processors and stuff that we’ve built internally with MixGenius.”
This machine learning aspect is very important, as MixGenius says that with each track mastered, and as the sonic diversity of those tracks grows, LANDR’s processes get more finely tuned. Since the first iteration of LANDR launched about a year ago, it’s mastered over one million tracks.
Currently, the main form of user control in the paid version is indicating “where you sit in the loudness wars” with the Intensity setting. Evans says the MixGenius team, which includes a number of mix engineers, has been wary about giving users too much control over settings. “Because everything is done algorithmically and on a frequency analysis basis, there’s a lot of stuff we can do really, really well, and it’s the science side of mastering. If we give people a whole bunch of options, then it’s going to potentially ruin that,” he says.
“We’re changing the workflow for people and I’m really excited about that,” says Evans. “People are using the same track over and over again, which is a use we didn’t anticipate at all. Because it’s so inexpensive, people will dump their tracks in, remix it, and go, ‘Oh, that’s what it’s going to sound like mastered’ and then they take it to the mastering studio for that last 10 per cent that we’re not achieving for them, but in the meantime, they’ve mastered their song 20 times and have a way, way easier workflow with the mastering studio at the end. They understand what they want with the master and are more articulate with it.”
The User**Patrick Gregoire, musician/producer/engineer/label owner**
Patrick Gregoire first gained wide attention as a member of the now-disbanded group Sister Suvi. He is now the driving force behind the group Pat Jordache, produces and mixes for a number of artists, and has recently built a hybrid digital/analog studio and mix room in the Hotel2Tango in Montreal.
“I don’t necessarily use LANDR for my clients when they are going to be releasing something. I recommend that if someone has the money, it’s always better to have a human-supervised mastering job. But what I do find it’s been useful for, and what a lot of people have been using it for, is as a reference opinion,” explains Gregoire. “It’s this place where you can send your music and at least get some rough idea of what it’s going to sound like after a mastering job – that you can use as a benchmark, essentially. Often what I’ll do once I’m done with a mix and I don’t really know where to take it from there is I’ll LANDR it and see what they do to it. It is kind of a way to quickly identify any problem that may be lingering.
He continues, “Then I’ll either go back into the mix and tweak how I’ve been working on things or even do my own mastering job and try to beat LANDR. That’s actually kind of one of the fun things, where you A/B back and forth until you feel like you can actually get better than what they’ve done.”
One other use for Gregoire, since he often performs solo with some instrumental backing tracks, is, “in a pinch,” he’ll put a mix for an upcoming live show through LANDR to get a “uniform loudness and consistency of return on it.”
“I know a lot of engineers that are sceptical about it or feel possibly threatened. I think that [MixGenius] have done a good job of representing themselves as not something that’s going to be necessarily replacing human mastering, but that can fill all these other functions that are starting to make themselves more and more apparent. People don’t always have the budget or time to go to a pro mastering studio and that when you’re just trying to get something out there quickly and efficiently, that this is something that is available for you.”
The Critic**Noah Mintz, Senior Mastering Engineer at Lacquer Channel Mastering**
Noah Mintz is considered one of Canada’s top mastering engineers, having put the finishing touches on albums by Arkells, Broken Social Scene, Death from Above 1979, and Rheostatics, just to name a few. When he saw an early press release about LANDR, he dismissed it as “ludicrous,” but decided to test out the free version as he began to hear about it more and more. When it comes to his views of LANDR, he doesn’t pull any punches.
“First off, you’re trying to turn an art into an algorithm, which is literally impossible. It just can’t be done. You can’t have a computer do mastering any more than you can have a computer create art,” says Mintz. “They can claim whatever they want, but as a 17-year veteran in mastering, much of what I do is in what I don’t do. It’s not the gain that I use or the EQ I use or the compression I use; it’s how I use the compression or why.” Because, Mintz says, mastering is all about maintaining or enhancing the listening experience and artistic intent, not necessarily the technical quality, he sometimes makes no changes to a mix, but I’ll take him eight hours to come to that conclusion.
“I don’t understand how an algorithm can listen to a file. It can analyze it, but it can’t listen to it. It doesn’t understand how a set of harmonics or played frequencies touch a human being, right? So it can’t make those decisions that are important towards a quality listening experience, which is really what all mastering is about.”
Taking on the notion that LANDR is “good in a pinch,” he says, “We have enough things out there that are contributing to the degradation of sound quality and I don’t necessarily think we need another.”
Mintz does note that his opinions don’t stem from a place of competitiveness. “There is probably a very small percentage of releases out there that actually use professional mastering, so LANDR is not really competing with us; it’s competing with smaller-grade mastering or the studio mastering, and those guys can do a better of a job than LANDR can do,” says Mintz. He also takes exception to the notion of machine learning. As he says, mastering one million tracks makes no difference to a robot unless there is a control in place telling LANDR what it’s doing wrong and what to do differently next time.
“I am the first person to adopt technology; I love technology of all sorts, and if I really felt what engineers do could be quantified or imported into an algorithm, then I wouldn’t have a problem with it,” Mintz says in conclusion. “You can look at the RMS of a song, raise that level, and maybe that is good in a pinch, if that is what [LANDR] really does, but anybody can do that who has recorded an album.”
*Michael Raine is the Assistant Editor of *Canadian Musician