You are here

Sounding Off: Non-AI Mixing

Jayne Drake
Published November 2017
By Jayne Drake

Will non-AI mixing become the new vintage?

The year is 2030 and our musical lives are easier. Composing and producing doesn’t require us to be an expert in loudness units, know when to use an optical compressor or worry about gain staging. Intelligent DAWs have learned how to automatically mix our music using petabytes of cloud data, harvested from the best mix engineers around the world. Algorithms from the best mix engineers can instantly analyse the spectral content within our own recordings and offer world-class mixes at the click of a mouse.

Sounding OffAll our previous interactions with our DAW and online behaviour has provided increasingly targeted mix algorithms that echo our own musical style. The music we stream in our cars, homes and all of our browsing habits provide a rich data set for our AI service provider. Now it’s all about our musical ideas, arrangements and performance — which AI-enabled DAWs can also help with, by suggesting ideas based upon our past musical behaviour, digital footprint and unique online identity. Years of personal data harvested from our Internet activities provide a rich personal profile, matched to the best mixing algorithms in the cloud.

Musicians smile about earlier generations discussing analogue or in-the-box mixing techniques. These days music can be recorded effortlessly to the cloud through everyday devices, including fashionable clothes, which seem to be everywhere.

Buying software is a strange concept. Why buy? Services are free or only require a micro-payment per mix algorithm. Our AI service provider knows how much we’re willing to spend (or not), so only offers us mix algorithms from around the world that match that our budget. Why bother buying plug-ins and get into behind-the-scenes technology? Instead, spend the money on the best algorithms you can afford. The absolute best ones do tend to cost a bit more, although many are free, as long as we trade personal data from our digital footprint. Our personal data helps different AI service providers offer more targeted adverts.

So why is it that producers hanker for old, difficult and expensive ways of mixing with plug-ins and hard-to-learn software? Lots of reasons. We’re human and like to be in control. Music is about our own identity rather than someone else’s. We love the challenge of the ‘new’ rather than recycling the work of other humans or machines. Producing the finished product is more satisfying and, anyway, we can offer our own mix algorithms to others, so we build up our own online profile in an industry saturated by aspiring composers. In a world where everything is done for us — including our thinking — we need music to call our own. Who needs machine-learned skills from the cloud when we have our own?

Automated data mining and analytics from the big AI service providers have spotted a trend for individual mixing solutions rather than those distributed through the cloud. There’s an emerging market for old-school, non-AI music production, off grid. This emerging user community has chosen for themselves who and what they love: music created and mixed by people, for people. Not the machines.

Younger generations seem happy to simply pick up an acoustic instrument and focus on what they do best and let the AI take the mixing strain. Non-AI mixing seems strange to younger musicians who leave decisions around parallel compression and side-chaining to their DAW and the older generation. Younger generations just want to make sure others hear their music online — which is a real problem given the sheer volume of automated music production that competes with traditional performers. Mainstream radio and television stream content that matches people’s digital identities, rather than broadcasting the same content to everyone, which only encourages musicians to follow existing genres to maximise their chances of being heard.

Back to 2017... It’s obviously impossible to know what the future really holds and whether AI, machine learning, algorithmic mixing and big data will change how we work, or want to work. Analysing mix stems to generate a world-class mix would be a huge computational task given the thousands of subjective decisions made within a typical production. But trends in music technology have sometimes been surprising in the past, so perhaps non-AI mixing will become the new vintage. Who knows?

About The Author

Jayne is a music producer and a director at Beyond Radio. She also finds the time to hold down a normal job in IT.

Published November 2017