Neural-vocal phase coupling reveals structured timing in birdsong production

Understanding how neural activity encodes motor behavior during vocal production is an unresolved challenge in neuroscience. Songbirds provide an ideal model due to their song dedicated neural networks. Here, we analyze songs from adult canaries (Serinus canaria) by comparing multiunit neural activity recorded in the telencephalic nucleus HVC (MUA) with the acoustic envelope of the produced song (ENV), aiming to identify temporal delays reflecting neural–vocal coupling. We introduce an approach based on cross-correlation functions in short sliding windows, using surrogate data for statistical validation. The resulting delays (τ) are grouped into populations using kernel density estimation, where local maxima and valleys define objective boundaries. While canary song is known to be stereotyped and rhythmic, we find that neural activity also displays a rhythmic organization, temporally aligned with vocal behavior. We identify regimes across birds, including synchronization near τ = 0, positive and negative delays, under correlation and anticorrelation conditions. These regimes reveal multiple structured temporal relationships between neural activity and vocal output, consistent with the known sensorimotor role of HVC, where neural dynamics span distinct phases relative to behavior and may reflect different contributions to vocal production.