Roles of frequency-modulated components of monosyllabic sounds in auditory processing

Noise burst (NB), frequency-modulated (FM) and constant frequency (CF) components are considered major vocal components of monosyllabic sounds. Based on hierarchical neural network architecture, we investigated roles of frequency-modulated components in monosyllabic sound processing. The neural network of the first stage decomposes monosyllabic sounds into time varying spectral components (NBs, FMs, CFs). Three neural networks of the second stage detect these components and encode them as dynamical point attractors of network dynamic, which then are integrated by the neural network of the third stage into single point attractors corresponding to the monosyllables. Perception of a monosyllabic sound is the emergence of the single point attractor at the third stage corresponding to the monosyllable. We demonstrate that frequency-modulated components are advantageous for encoding monosyllables into relevant dynamical point attractors but not necessary for monosyllabic sound perception.