MLNI Week 2:ML/AI Case Study – Alex Wang

AI Technology in music production

Intro:

lately I got really interested in music production, and I realized that there are many new technologies available for music production that is actually powered by machine learning! 

Phases of music production:

There are multiple phases in the production of a song. A composer writes out a song, a producer then creates it using a digital audio workspace, then a audio engineer mix and masters the track to perfect its dynamics and clarity.

The role of an audio engineer:

Audio engineer actually plays a really big role in the making of a song, it is not required for them to have any knowledge in song structure, music theory , instrumental skills, etc. But they are the ones that makes the song sounds perfect without creating anything themselves. A audio engineer can spend his whole life perfecting his skills as it is a very complicated job.

AI as audio engineer:

AI is very good at replacing humans in areas that does not require creativity, but requires skills. So music mastering is actually what people have been working on to let AI replace. You do not need to create anything, but you need very good ears and experience to perfect the song. 

I was at a speech by Kai fu Lee here in NYUSH a year ago, and he talks about the types of jobs that are easily taken over by AI and the ones that needs more human elements to it. Audio engineers, even though they are really respected and takes years to optimize their skills. They are still a job that does not require creativity and human compassion, which is why they are currently a target to be replaced by technology.

Landr/Ozone:

existing services like Landr and Ozone is already using AI as a tool to master the track for you. These convenient services, along with the advancements in making DAW and samples more accessible, it is actually really easy to get started in making professional sounding music nowadays.

https://www.youtube.com/watch?v=43Uad9C6LeQ

Articles:

WILL AI REVOLUTIONIZE THE WAY ARTISTS MAKE MUSIC AND FIND SAMPLES? LANDR SAYS YES…

Landr raises $26 million for AI-powered music creation tools

“Original” work with python Neural composer + Izotope Ozone mastering:

I made a song using melodies composed by a python network, I did the producer part of the creation process, while Ozone AI did the mastering of the track.

Reflection:

Over the past few years I realized how AI is capable of not only replacing human in physical labor, but also be an amazing resource in the world of arts. Even though they might not yet be capable of replacing artists, they are now a very strong tool in the assistance of making quality products. They are closing the gaps between a professional musician with a expensive studio and a whole team of producers/audio engineers along with professional gear, and any ordinary person with a laptop.

Leave a Reply