The Centre for Vision Speech and Signal Processing is offering a PhD studentship to pursue research in speech and audio technologies, which include speech coding, speech analysis, speech synthesis and modification, audio classification and retrieval, speaker animation, and automatic speech recognition. Sound is a medium that most people use on a daily basis for communication and interaction, and this research is concerned with seeking to improve and enhance those experiences. So, these technologies are of particular interest to companies and organizations involved in broadcasting, computer games, language training, or helping those with speech and hearing disabilities. The research is equally concerned with understanding the features of sounds such as speech that are important for perception. The long-term aim is to develop machines that can interact with us naturally and intelligently.
The studentship includes home tuition fees and an annual bursary of £9K (depending on qualifications and experience), suitable for a UK or EU citizen. The successful applicant will join an expanding research group within the Centre for Vision Speech and Signal Processing, which currently has three professors, eight lecturers, a dozen RAs and about fifty PhD students. CVSSP has an international reputation for the excellence of its research and, in the last Research Assessment Exercise, the Department (of Electronic Engineering) was awarded grade 5*A.
The studentship is to undertake research in the area described below (though other proposals in speech and audio processing may be considered). Students are expected to contribute to, and benefit from, existing research activity. For information about current research at Surrey in this and related areas, click on www.ee.surrey.ac.uk/CVSSP/.
Candidates should hold a 1st or 2i honours degree or Masters degree in Electronic Engineering or related discipline (e.g., Engineering, Physics, Mathematics or Computing), and should be able to:
Automatic speech recognition systems generally treat all sounds alike, whether they were spoken, sung or whispered. However, sounds like /s/ and /sh/ have characteristics that are very different from those of vowels, say. Indeed, for certain sounds, like /z/ or /zh/ (as in measure), part of the sound is noise-like while the other part is almost periodic. Using a technique to separate the periodic and aperiodic components of a speech signal automatically, this project will examine the role of differing aspects of sound quality to achieve improvements in speech recognition accuracy. Preliminary findings in collaboration with researchers from the Universitat Politècnica de Catalunya in Barcelona have been promising, but the application of acoustic segregation methods for this purpose remains largely unexplored.
© 2002, maintained by Philip Jackson, last updated on 26 November 2002.