We used a simple generative model to synthesize novel sounds with naturalistic properties. We found that such sounds could be segregated and identified if they occurred more than once across different mixtures, even when the same sounds were impossible to segregate in single mixtures.
Page 1 / 1
