16. Music AI#
In this chapter, we explore the developing intersection of music and artificial intelligence (AI). Recently, it has become possible for anyone to generate realistic music at the push of a button, in seconds. For example, here is what a commercial music AI system, Suno, produced from a single sentence of text:
A song generated by Suno (v6) from the prompt “disco funk track about the history of computer music”. Everything, from the lyrics to the vocals to the arrangement, was generated.
The generated lyrics (first verse and chorus)
Bell Labs, nineteen fifty-seven / A room of machines and a note from heaven / Max Mathews wrote the code to make / A computer sing for the first time, awake / Numbers became a tone in the air / A new kind of instrument started there
Turn it up, let the circuits play / From the first small tone to the sounds today / Hands on keys or a model’s reply / We keep making music from the why and the try
How did we get to generation of realistic music audio from the simple Markov chains that we studied in Chapter 14? What are the implications of this technology for music, creatively, economically, and culturally? This chapter offers a brief tour of the first question, and offers some perspective on the second.