Automatic speech transcription in AAL solutions

Voice is a valuable channel for extracting information on user's context, intentions and needs within an AAL solution. Automatic speech transcription (speech-to-text) is an inevitable step in semantic analysis of speech content. In this paper we consider challenges, limitations and perspectives of speech transcription application in AAL solutions based on and generalizing the experience gained in HERMES EU FP7 project. © ICT&S Center, University of Salzburg.