Parallels and divergences in spoken word recognition between humans and a neural network model

Recent advances in artificial neural networks have enabled the design of automatic speech recognition systems that perform comparably to human listeners in various speech recognition tasks. Careful analysis of the behavioural characteristics of such systems can reveal similarities and critical divergences between machine and human speech recognition. These insights may inspire testable hypotheses using purpose-built artificial neural networks, human psycholinguistic or neuroimaging experiments, and eventually further our understanding of speech perception. We used a recently published end-to-end model of human speech recognition and compared its behavioural characteristics to two aspects of human spoken word recognition; investigating whether the model exhibits similar behaviour when adapting to changing speech rates and speakers. We found that the network model responds similarly to humans to speech rate changes, but not to speaker changes. We are investigating which architectural features could explain the lack of speaker change effects and plan follow-up simulation and behavioural experiments.