Birds have swing: Multifractal analysis reveals expressive rhythm in birdsong

Abstract

Music is thought to engage its listeners by driving feelings of surprise, tension, and relief through a dynamic mixture of predictable and unpredictable patterns, a property summarized here as "expressiveness". Birdsong shares with music the goal to attract and hold its listeners' attention and might make use of similar strategies to achieve this goal. We here tested a songbird's rhythm, as represented by the amplitude envelope (containing information on note timing, duration, and intensity), for expressiveness. We used multifractal analysis which is designed to detect in a signal dynamic fluctuations between predictable and unpredictable states on multiple timescales (e.g. notes, subphrases, songs). Results show that the songs' amplitude envelope is strongly multifractal, indicating rhythm is patterned by fluctuations between predictable and unpredictable patterns. Moreover, comparing original songs with re-synthesized songs that lack all subtle deviations from the "standard" note envelopes, we find that deviations in note intensity and duration significantly contribute to the rhythm's multifractality. This suggests that birds render their songs more expressive by subtly modifying note timing patterns, similar to musical operations like accelerando or ritardando. Our findings bear consequences for neuronal models of vocal sequence generation in birds, as they require non-local rules to generate rhythm.