AI in Music

I vividly remember when Apple first released GarageBand for macOS in 2004. It was a free music production software application that allows anyone to create music and on a Mac. In 2011, versions for the iPad and iPhone were added. As a classically trained musician and former instrumental music educator, I was blown away with the kind of music-making you could do with this app. Our twin boys were only six at the time and they thought it was so cool to “make up music” that sounded fun. Here’s one of their favorites: “The RV Song.”

MidwestCon

Fast forward to September 2024, twenty years later, I was at MidwestCon, a policy innovation conference for tech and AI curious, builders, beginners, policy advocates, entrepreneurs, inspiring entrepreneurs, designers, creators, marketing and branding professionals, blockchain builders, developers, students and educators. Although interestingly enough there were very few educators in attendance. Strange I thought, given the impact AI is having, and certainly will have, on education.

Kaila Love

One of the breakout sessions I sat in on was presented by the hip hop artist and AI automation specialist Kaila Love. During her breakout session she had people break into small groups and make up pop songs on their phones using Udio and Suno It was very audience-centered with lots of active-participation. Suddenly, I  had a flashback moment as I was once again blown away with the kind of music-making that was happening. Even though most of the participants were not trained musicians, there music creating talents using these apps was impressively cool.

And in the hands of musicians (or musically informed AI prompt engineers) listen to what’s possible out “in the wild” with Suno:

BASSA NOVA
Bossa Jazz A Cappella by 3Daizy

MEMPHIS SOUL
Ain’t Got a Nickel Ain’t Got a Dime by Soul Diego

POP
Stone by ImOliver

METAL
Second Law of Thermodynamics by Stei Camel

Copyright Concerns

Having been involved in the music publishing business for a hot second or two, it was interesting to think about all the music copyright considerations in using these apps. One common complaint from the “breakout session creators” was when they asked the app to “sound like” the Beatles or Mariah Carey, they were informed by the app that they couldn’t do that? Perhaps some proactive lawsuit prevention?

Generative AI

Never thought that our twin boys were utilizing AI back in 2004. Yet, there was some element of guided predetermined musical intelligence that was able to assist them in “making music.” I mean think about it. Generative AI uses patterns learned from existing data to create entirely new content, by identifying and replicating the underlying structures within that data to produce original outputs that seem similar to the training data. Garage Band also uses patterns learned from existing music to create entirely new music by identifying and replicating the underlying structures and form within existing music to produce “original music” that sounds like existing musical genres and styles.

I realize GarageBand is simply a digital audio workstation (DAW) that allows users to record and play back multiple audio tracks. It also includes built-in audio filters that can add effects like reverb, echo, and distortion. It even has virtual drummers, like “Leah” and “Jasper,” users can integrate within compositions. Apple claims these virtual drummers can play millions of different combinations. Who needs a drummer?!

Spotify

Soon, you might not even need a band to make music!? In April 2024, Spotify introduced AI Playlist into beta, in the U.K. and Australia. It started as a tool to fuel smarter automagical discovery and playlist curation. In September of 2024 the feature offers users additional opportunities to “express themselves through the music they love” adding the beta in the U.S., Canada, Ireland, and New Zealand.

Beyond curating playlists, Spotify is reportedly adding AI generated songs to stream alongside real-world performing artists. Jet Fuel & Ginger Ales is a band that has earned the “Verified Artist” badge on Spotify with 414,500 monthly listeners for its covers of The La’s and Red Hot Chili Peppers. Yet, many are convinced that the band is AI-generated. “Other groups like Awake Past 3 and Gutter Grinders have hundreds of thousands of monthly listeners on the app and often share some suspicious traits: the uncanny valley vocals, the plain-Jane branding of their logo, and a dearth of biographical information.”

Scary to think that while Spotify fails to compensate real-world artists adequately, AI artists don’t get paid at all?! Subscribers are pushing back a bit, but who knows how this plays out.

Ironically, Spotify removes tens of thousands of AI generated songs that users create. Yet, when Spotify creates AI generated tunes it’s okay?!

Better, Faster, Cheaper

“Better, faster, cheaper—pick two” is a saying known as the “iron triangle” or sometimes called the “triple constraint triangle.” This saying supposedly originated with NASA in the 1990s when the agency launched a Faster, Better, Cheaper (FBC) initiative. The goal was to fly more missions, in less time, while reducing costs and increasing innovation. Well, as Dr. Phil would ask: “And how’s that workin’ for ya?” When it comes to AI in music, education, and many areas of our lives, “faster and cheaper” is already true. “Better” is still evolving into creating content that many just might be “good enough” for most.

AI Is Already All Around Us

Truth be told, AI is already all around us and whether we want to admit it or not, we all benefit—and are at risk from—the use of AI. The lessons I learned at MidwestCon are two very important ones:

#1. AI is moving faster than anyone cares to admit—even those “in the know.”
#2. AI has already permeated our lives more than anyone cares to admit.

Consider these examples of AI that already impact our world…

• Chatbots: Simple AI-powered chatbots have been used in customer service for years to answer frequently asked questions.
• Computer vision in agriculture: Computer vision technology, often powered by AI, allows farmers to monitor the “digital life cycle” of their crops by analyzing images captured through drones and tractors, enabling precise identification of areas that need specific treatments, thus minimizing the need for unnecessary pesticides and fertilizers, leading to a more sustainable and efficient farming approach— this is known as precision agriculture.
• Computer vision in manufacturing: Industrial robots utilize computer vision with AI to identify and inspect products on assembly lines.
• Email filtering: AI algorithms can identify and automatically sort emails into different categories like “important” or “promotions.”
• Facial recognition: Facial recognition software, used in security systems and social media platforms, has been developed over many years to identify individuals based on their facial features.
• Fraud detection systems: Thank goodness financial institutions utilize AI to analyze transaction patterns and identify potentially fraudulent activity.
• Medical diagnosis support: AI-powered tools can assist doctors in analyzing medical images and suggesting potential diagnoses.
• Music streaming suggestions: Music services like Spotify use AI to generate personalized playlists based on listening habits.
• Navigation apps: GPS navigation apps like Apple Maps and Google Maps leverage AI to provide optimal routes based on traffic conditions and user data.
• Personalized advertising: Online advertising platforms beyond just Google and Facebook use AI to target users with ads based on their demographics and interests.
• Plagiarism Checkers: Grammarly, Turnitin, Scribbr, PlagAware, Compilatio, and many others use AI to determine originality of content. (Although I wonder if we’re just using student writing to train more AI?!)
• Recommendations: Online platforms like Netflix and Amazon use AI to suggest products or content based on user history and preferences.
• Robot vacuums: Robotic vacuum cleaners use AI to navigate around obstacles and clean floors autonomously.
• Search engine algorithms: Search engines like Google use AI to understand search queries and provide relevant results based on complex ranking algorithms.
• Spam filters: Email providers use AI to identify and filter spam messages based on patterns in the content and sender information.
• Spellcheck and grammar checkers: Word processors utilize AI to suggest corrections for spelling and grammar errors.
• Targeted TV Advertising: Connected TV & OTT ad platforms provide advertisers with household specific targeting options to create direct-response TV advertising via Hulu, Spectrum, Comcast, and many others.
• Voice recognition software: Even way back to early versions of voice recognition technology like Dragon NaturallySpeaking (initially released in June 1997), this early AI allowed users to dictate text using voice commands.

Funny to think that as I type this my next and final conference trip of the year will be to The Midwest International Band & Orchestra Clinic. An event I first spoke at 30 years ago! And a place that “strengthens instrumental music education through extensive professional development, inspirational musical experiences, and the cultivation of rewarding professional relationships” in the real-world of instrumental musicians “making music.”