In order to begin to understand the neurophysiological basis of vocal learning, one must have analytic tools for understanding and characterizing complex vocally- generated, acoustic signals. With these tools in hand, one will immediately be better equipped to address a host of neuroethological questions about how vocal learning occurs. In our first aim we propose to develop and build these tools using the concepts and methods of Shannon's Information Theory. In our second aim we will use these tools to answer questions about how zebra finch song varies within and between normal adult individuals who are exhibiting either directed (singing to a female) or undirected (singing alone) singing behavior. In our third aim, we ask how similar are bird songs to the songs of their tutors. Knowledge of song similarities and differences in these populations will constrain and inform possible underlying neurological mechanisms of song production and learning. This knowledge is likely to be of general relevance in the study of disorders that affect human speech acquisition and learning. Furthermore, we are hopeful that our algorithms and methods of analysis will be useful in solving difficult problems in human speech recognition.
