Fonema

Context and constraints

Recorded voices pass through a neural network built from voice classification and recognition models. What comes out keeps the grain and rhythm of speech, but the words stop making sense.

What I built

• Medium: audio installation; recorded voices processed by a neural network [needs: channel count, duration, speakers]
• Shown: thesis project [needs: where shown; confirm it was the Conservatorio thesis]

Outcome

[needs: exhibition record]

Media

[needs: 3 to 5 stills, duration and excerpt]

[needs: streamable audio]

Links