Noise Reduction For mobile Communication

Improving QoS for Mobile Communication in Noise

EPSRC De-noise Project on Speech Enhancement in Noise Integrating Noise and Unified Speech Models for Communication on Mobile Devices in Noisy Environments (EPSRC no. GR/S30238/01)

Improving the Quality and Intelligibility of Speech From Cellular Phones in Moving Car/Train (EPSRC no. GR/M27968/02)

Demos of Speech Enhancement in Mobile Environment

Demo of Bandwidth Extension

Demo of Kalman filters

Accent Profile and Accent Morphing

Synthesis of British, American and Australian Accents

Accent profiles model and map the differences in the acoustic (formants and pitch trajectories) and phonetics of accents. (EPSRC no. GR/M98036/01)

Applications: accent synthesis, accent identification and accent-adaptive speech recognition.

Demos of Accent Synthesis

ARCHIVED Audio Restoration

De-noising of old archived records

Model-based Spectrum Extrapolation and Reconstruction

Demo of Narrowband to Wideband Speech Enhancement, Bandwidth Extension

Demos of Audio Restoration

Florence Nighingale, Mozaffareddin Shah...

Speaker Profile and Voice Morphing

Speaker profiles model all the acoustics, prosody, and delivery style of a speaker. (EPSRC no. GR/M98036/01)

Applications include voice morphing, speaker identification speech recognition and TTS.

Demos of Voice Morphing

IMAGE DE-NOISING, MEDICAL SIGNAL PROCESSING, MIMO

INDEPENDENT AND PRINCIPAL COMPONENT ANALYSIS

The aim of these projects are to develop independent component analysis (ICA) and principal component analysis (PCA) methods for image de-noising, multiple-input-multiple output (MIMO) medical signal processing applications and for eigen-analysis of bio-medical images.

Voice Quality Types & Glottal Flow Model

Glottal pulse shape influence voice type and identity and are modelled through dynamic decoupling of formants.

Applications include speaker identification/authentication and speaker dependent synthesis and recognition, voice type models.

Demos of Voice Types and Emotion

Multimedia Signal Processing Projects

Acoustic Echo Cancellation

Medical Image Processing

Signal Processing in Pollution Control

Restoration of Gramophone Records

Text to Speech Synthesis

Research includes selection of best concatenative units, synthesis of 'unseen' units and contexts, spectral smoothing across formant tracks, temporal models, prosody models.

Applications include computer voice, mobile devices, voice-mail, and in-car navigation systems.