57:16Microbes, Robots, and Ambition - Robin Sloan on His Novel Sourdough
From Y Combinator · Published Jul 22, 2018 · Watch on YouTube
TL;DR
Robin Sloan explains how he used a machine learning system (SampleRNN) to generate fictional alien music for the audiobook of his novel Sourdough, trained on Croatian klapa singing. He discusses the challenges of robot arms in kitchens (the “egg problem”), argues against framing old and new as adversarial (advocating for “and”), and describes his practice of deleting tweets to preserve a café-conv
Key insights
- Audiobooks are a huge and growing market, and producers now expect more than a simple recitation of the printed text.
- Sloan used SampleRNN on a corpus of klapa MP3s (Croatian folk singing) to generate the fictional music described in Sourdough; the model’s “failure” to perfectly replicate the input is what made the o
- He sees a parallel between his own creative process (listening → mashing → abstracting) and what the neural network does.
Want the full analysis - every claim cited to the second it was said?
This page only shows a teaser. Sign up to chat with the complete, cited breakdown of "Microbes, Robots, and Ambition - Robin Sloan on His Novel Sourdough".