Looking ahead: They came first for our art, then for our prose and jumbled essays. They now visit for music,

Looking ahead: They came first for our art, then for our prose and jumbled essays. They now visit for music,

With a "new" machine learning method, new music clips and genres may be generated, interpolated, and looped via picture creation.

With a "new" machine learning method, new music clips and genres may be generated, interpolated, and looped via picture creation.

The Stable Diffusion (SD) algorithm was modified for music by Seth Forsgren and Hayk Martiros, resulting in an unusual new kind of "music machine." 

The Stable Diffusion (SD) algorithm was modified for music by Seth Forsgren and Hayk Martiros, resulting in an unusual new kind of "music machine." 

Riffusion, like SD, turns a text input into original, AI-generated amusement. The main difference is that sonograms were specifically used in the system's training.

Riffusion, like SD, turns a text input into original, AI-generated amusement. The main difference is that sonograms were specifically used in the system's training.

As explained on the Riffusion website, a sonogram (or a spectrogram for audio frequencies) is a visual way to represent the frequency content of a sound clip.

As explained on the Riffusion website, a sonogram (or a spectrogram for audio frequencies) is a visual way to represent the frequency content of a sound clip.

Time is represented by the X-axis, while frequency is represented by the Y-axis. Each pixel's colour indicates the audio's amplitude at the frequency and time specified by its row and column.

Time is represented by the X-axis, while frequency is represented by the Y-axis. Each pixel's colour indicates the audio's amplitude at the frequency and time specified by its row and column.

Time is represented by the X-axis, while frequency is represented by the Y-axis. Each pixel's colour indicates the audio's amplitude at the frequency and time specified by its row and column.

Time is represented by the X-axis, while frequency is represented by the Y-axis. Each pixel's colour indicates the audio's amplitude at the frequency and time specified by its row and column.