Can robots really master our music ? Kenny Gioia set up poll on Facebook to see, asking people to choose between six different masters of the same song - three by humans (John Longley, an anonymous engineer from Sterling Sound and Steven Slate) and three by machines (LANDR, Aria and iZoTope Ozone).
The great thing is, the test was blind, and loudness matched - so no-one knew which master they were listening to, and loudness didn’t influence the way the masters sounded.
And the results of the poll were fascinating. In this show Kenny reveals which master was which, and we discuss the results - including:
- Why this is not a scientific test
- Ian’s blind comments on the masters, including his instincts about which was which
- Why the poll results are probably skewed
- When automated makes sense
- What the robots still can’t do - and maybe never will
Links
Kenny’s Groove 3 Gurus
Download the files
See the poll
Ian’s LANDR test
Graham’s shootout - Ian v. LANDR v. a limiter
Ian’s Perception Plugin
Startup Podcast - Driverless cars
What happened to #42? 😉
long story but we're doing #42 next and hopefully it will make sense why then. It's Ian's idea.
Aaaah, right. Thought you’d both suffered from some brain cell loss and totally deleted #42 from your memories 😉
Great show, as ever, really appreciate what you guys do - Thank you
Great outlook with the ideas expressed with man vs automated machine for mastering. Question for future episode chat.
What are your thoughts with mixing your tracks of a song to pink noise?
One of the things that is prevalent these days is the allure of new technology as a replacement for humans. We see it everywhere with instant checkout at supermarkets to contactless payment.
A couple of factors not mentioned in the show and that may have an influence on who uses automatised mastering, are cost and human relationships. The cost of hiring a human is a major factor with indie artists, particularly those that are just starting up. More people are turning to home studio methods to produce these days and the cost is minimal for top quality end results.
Human relations is a huge topic but briefly speaking, we are loosing the art of social interaction more as we hide behind avatars and the like. From an artists point of view it is particularly hard to lay months/years of mixing in front of a complete stranger for mastering.
It’s even more complex than that when we filter in the fact that very few understand the job of a mastering engineer and even believe mastering is not needed.
Even if we accept mastering, most will find it very hard to find a mixing engineer that is empathetic enough to “get” the vibe of the music. Critical feedback is very hard to take, no matter how well meaning, particularly when there are artistic differences between artists/mixers and mastering engineers.
When we factor in all the above and more, it is easy to see that automated mastering is going to become a favoured option - very sad, as there is no substitute for the human experience.
As much as I apprechiate human handicraft of experienced persons, when you talk about the mastering services as machines, you have to remember it's human experience embedded into code. And just as Jeopardy champions are beated by the Watson AI program, automated mastering will eventually be prefered by a majority listerners. It's about feeding human experience into code. We are not there yet and probably not next year - but it's just a matter of time and effort... And the conversation you are looking for can be had via a UI. The trivial UI of LANDR e.g. is not the right way for sure.
Regards the hyping of bass, could 'Mega Bass' or such like settings on players be having an influence. Are people getting used to a hyped up bass, is that what they now think is normal?
Thank you all for your input in todays audio world,.... i think it's a bit dangerous to generalize gain fluctuations (post master), since in fact end-listeners play their gains relative to their soundsystem and would potentially change gain decisions through a different system. from my 30year experience i have found that i must be specific on each system.... be well all, kenny's comment on novices and automated mastering is very poigniant!
Thank you, Jon, for finally mentioning near the end of the show what I was thinking throughout: What about album mastering? For me, the real value, and art of mastering is making the songs of an album fit together. If automated mastering ever attempts to pull that off, I'll be much more interested, but still skeptical. (I don't master much. Nothing critical. But I still learn a lot from your podcasts. So much can be applied to mixing.)
Going thru the backlog of the podcast and got to this one. I'd LOVE to listen to files myself for comparison but unfortunately the original Facebook post's link to files is broken. Any chance anyone here (or Ian) has a working link to those files? Thank you!