Recently Udio, a groundbreaking text-to-music model, was released to the public. Following in the footsteps of prompt-to-text, prompt-to-image, and prompt-to-video technologies, text-to-music has now become a reality. Among the pioneers in this domain, UDIO and Suno stand out as the most capable text-to-music models currently available, and they’re both accessible for free.
The buzz after UDIO’s release, made me curious and I wanted to put its capabilities to the test. From a user’s standpoint, UDIO is simple to use. You simply provide a text prompt to generate music. UDIO allows you to use lyrics in a separate text prompt, thus incorporating them into the generated music. Each iteration of UDIO generates a snippet of the song, typically around 30 seconds in duration, accompanied by two alternate options for further development. To get a song structure, users can use tags like [intro], [verse], [chorus], or [outro].
Excited to explore UDIO’s potential, I asked it to generate several songs. One particularly inspired piece began with a text prompt requesting a “roots reggae protest song with a male vocalist.” Employing Chat GPT 4 to generate both the lyrics and the song’s structure, I drew inspiration from one of my favourite podcasts, “Philosophize This.”
The resulting song was impressive. Actually I think it is too good. It begs the question: what does it mean to be creative in a world where creativity itself can be automated? And furthermore, in such a landscape, what does it mean to be human?
Buying green ain’t enough: An AI Reggae Song
Listen on Udio
Lift the Shadows: An 80s AI song about populism
Listen on Udio