The area of speech technology includes a variety of applications where methods for digital processing of speech are crucial:

• Speech coding (compression),
• automatic speech recognition,
• speech synthesis and phrase concatenation,
• speech analysis,
• speaker verification and identification,
• speech based human-machine dialogue systems and
• speech enhancement.

Internationally speech processing has been a research area for more than 60 years. The ultimate goal has been machines that will understand and produce speech as naturally and easliy as humans. Using speech as a mean for communication is simple and natural for humans. Thus a logical goal is to use speech as an interface between humans and machines instead of keyboards, mouse, touch-sensitive screens etc. In addition to improving the usability of the interface for people in general, speech based interfaces will dramatically improve the life quality for many groups of diabled, e.g. people impaired with respect to movements, sight, hearing and talking. During the last decades there has been an enormous increase in access by telecommunication (speech over telephone). Speech coding (including standards) has been a major contribution to this development, and also to the introduction of digital mobile telephony (GSM etc.).

Speech processing has been an important area of research within our university group for more than 30 years. As the group is located within the Department of Telecommunications, speech coding for different telephony based applications was a natural main area of research. However, for the last 10-15 years, focus has been directed towards automatic speech recognition (and, to some extent, speaker verification and speech analysis).

Lately, the group's resarch within speech processing has been concentrated on the development on speech based systems for human-machine dialogues. This area requires, in addition to speech processing, also knowledge within phonetics, linguistics, artificial intelligence, data base queries, etc. As a consequence, a multidiciplinary cooperation within the university has been established.

The current main resarch activities within speech processing are concentrated within the following three different research projects: SPODIS (SPOken DIalog Systems for telephony), TABOR and KIKS. For the time being, 5 Ph.D. students/scholarships within speech processing is attached to these projects.

Relevant courses for graduate studies:

• 42123 Speech and Music Technology
• 42521 Information and Coding Theory
• SIE2025 Digital Signal Processing
• 42535 Digital Coding of Analog Signals

Relevant courses for Ph.D. studies:

• DIE0901 Pattern Recognition
• DIE2954 Statistical Signal Theory
• DIE2975 Spectral Estimation
• DIE2976 Speech Processing

Contact person: Torbjørn Svendsen