[machinemusic]

A warm thank you to the Hacker News community for all the conversation after the Outside Lands Festival, especially around how Tupac's voice was engineered for the Drake track. For those curious about the mechanics, here’s a breakdown of how a machine approaches this:

First, the machine analyzes the input by breaking the sound down into its core building blocks (spectral morphing) while also reading the text to map its emotional content. It then actively reshapes the audio by adjusting its rhythm and timing, creating an immersive soundscape, and adding nostalgic textures based on that emotion analysis. Finally, the system fine-tunes these changes and polishes the track to create the final, normalized audio output.

On a final note, I'll be at the AWS Builder's Loft in San Francisco today. Hope to see some of you there!

ps- we will not re-engineer Tupac's voice - thanks for sending your remixes!

https://memorysoundscapes.app/