Technology is improving by the day and a lot of ordinary things that used to be delegated to human beings are now being passed on to AI (artificial intelligence) which is simply intelligence that is displayed by machines and not human beings.
one of the questions that makes its rounds in music production communities is;
“Can AI Replace Audio Engineers?”
AI cannot completely replace audio engineers because it is not yet advanced enough to do so. Human beings are beings driven by various things outside of logic and this makes them suitable to work in various environments. AI is driven by logic and while this is important, it is only a small aspect of any form engineering.
Why it may difficult for AI to Replace Audio Engineers:
Music Perception
One of the reasons that make it quite a difficult task to have AI completely replace a human audio engineer is because of music perception.
Music perception is simply the way a human being perceives music and it is different for every human being.
This becomes very complicated if you have to design AI to be able to compute and work out human perception. You also run into the problem of the AI designer implanting their own audio perception into the design because it is all they basically perceive. Which could create a serious problem because the AI would perceive this audio perception as the end-all and be-all perception.
Seeing that perception of audio for every human being differs, the AI would have to figure out a way to apply itself in a way that would have it engineer audio to an extent to which it would be deemed perfect or near-perfect by every listener or atleast almost every listener.
Complexity of Audio Engineering
Audio engineering is very complex and requires a lot of effort and unique approaches to each and every mix.
Therefore designing AI would mean having to account for every single thing that you would need to have in consideration when engineering.
This makes it a very difficult task to carry out because essentially we are talking about completely phasing out the human audio engineer and replacing them with a machine that should be able to do everything that a human being was able to do.
Workable Sample Size
When it comes to designing the AI to be able to meet human varying perception levels, you’d have to feed it a lot of varying data for it to process and utilize in making decisions.
This becomes a complicated task because the sample size you would have to put together would have to be a large one, with each individual selection catering to a specific set of unique audio perception parameters.
Adaptation
One of the easier ways that could be utilized to deal with these issues discussed here would be adaptation.
Adaptation would have to be engineered into the AI so that it can be able to adapt to certain parameters when they are fed to it.
This can be a complex task that would require years and years of trial and error to even be atleast 70% accurate. This is one of the reasons why AI alone cannot be completely left to take over audio engineering.
Response
Audio engineering is highly dependent on how quickly and accurately an engineer can detect problems. It doesn’t end there, an engineer also needs to know how to implement various strategies to deal with these problems.
AI would have to be as responsive as human beings and they should be able to solve problems that they are able to pick up.
When the details are too specific it can be difficult to have an AI to be able to fully grasp them and be able to work as well as a human being would.
This basically brings me to my last point.
Human beings are hard to replicate
It doesn’t matter what sort of AI you design, the most difficult problem that you’ll encounter is the difficulty that comes with mimicking and replicating human behaviour.
This point is one that has shown up over and over in this article. While I feel AI can be developed for audio engineering, it would certainly not be a simple task at all.
