Machine-Based Decoding of Voices and Human Speech
Alessandro Vinciarelli · 2018
This chapter provides an overview of the main approaches adopted in computing science to automatically extract information from speech signals. In particular, the chapter shows how the adoption of statistical models allows one to automatically transcribe speech recordings; that is, to automatically map the speech signals into sequences of words expected to correspond to what the speakers are actually saying. Furthermore, the chapter introduces the methodologies aimed at detecting the physical traces of social and psychological phenomena in speech, in particular when it comes to non-verbal and paralinguistic aspects. In both domains, the chapter outlines the main technological issues that need to be addressed in order to effectively process the data.