A-Lab: Machine audition
The machine audition lab (A-Lab): A-Lab develops audio signal processing and pattern recognition technologies.
Past projects such as EPSRC Qestral have developed of mathematical models for objective assessment of perceived spatial audio quality in surround sound systems. Research has also addressed the modelling of fricatives in speech, exploring the fusion of vision (lip tracking) and audio (phoneme recognition) in order to recognise and speech more effectively. Speech synthesis for talking 3D heads has also been explored, fusing audio and visual technologies to create more realistic facial animations using captured face using captured face models.
Ongoing research explores these and other audio related topics including solutions to blind source separation (referred to as the Cocktail party problem), 3D spatial audio and perceptual models of audio.