Speech Prosody 2008
We speak to convey information to the listener, and we listen to decode information carried by the speech signal. How we are able to do so is the ultimate puzzle for speech research. Much of the existing research effort, however, is devoted not directly to this central puzzle, but to various what could be called epiphenomena: speech rhythm, prosodic hie hierarchy, rarchy, intonational structure, naturalness of synthetic speech, etc. In this paper I argue that cracking the central puzzle of speech coding is not only the ultimate call for us as speech scientists, but also the key to understanding various epiphenomena in speech. I will demonstrate that speech involves multi multi- dimensional information coding due to the richness of information to be encoded and the compl complex exity of the ity underlying neuro neuro-physiological and biophysical mechanisms. Understanding this process may l lead to ead better understanding of many of the epiphenomena as well.
Bibliographic reference. Xu, Yi (2008): "Multi-dimensional information coding in speech", In SP-2008, 17-26.