Babble Model Parameter Variations Sounds

These sounds were produced by a computational spiking neural network model of learning to produce syllabic vocalizations (i.e. canonical babbling) in infancy. The vocalizations produced by each simulation from each parameter combination were sampled at regular intervals. For each simulation, these samples were concatenated into a single sound file that can provide an auditory sense of how the simulation's vocalizations changed over time. The best parameter combination is 200 neurons (200n) and 2 muscle scaling (2n).

本数据集的音频由用于模拟婴儿期学习音节发声（即标准咿呀学语，canonical babbling）的计算脉冲神经网络（spiking neural network）模型生成。针对每一组参数组合开展的每一次仿真所生成的发声信号，均以固定间隔进行采样。针对每一次仿真，所有采样信号会被拼接为单个音频文件，可直观展现该仿真中发声信号随时间的变化过程。最优参数组合为200个神经元（200n）与2倍肌肉缩放系数（2n）。