Dr Peter Jancovic

Lecturer,

Multimodal Interaction Laboratory

School of Electronic, Electrical and Computer Engineering, The University of Birmingham, Edgbaston, B15 2TT Birmingham, UK

Room: N518

Tel: +44 (0)121 414 4316

E-mail: p.jancovic@bham.ac.uk


NEWS

RESEARCH

TEACHING

LINKS

 

 

Research Interests

  • Digital Signal Processing
  • Speech Pattern Processing: Automatic Speech and Speaker Recognition
  • Noise Robustness in Speech Pattern Processing
  • Speech Signal Enhancement
  • Independent Component Analysis
  • Pattern Recognition
  • Image Processing

Teaching (academic year 2011/12)

EE2F1 Speech and Audio Technology - Undergraduates Year 2 course

Spoken Language Processing - MEng (year 4) and MSc course

Embedded DSP - MSc course

IET Competition 2010 - Photos

News (June 2009)

PhD Opportunity
School of Electronic, Electrical and Computer Engineering
The University of Birmingham
Automatic Recognition of Bird Species from Birdsong

The School of Electronic, Electrical and Computer Engineering at the University of Birmingham has funding for a PhD project in the automatic recognition of bird species from their song. The project will be undertaken in collaboration with Biocensus, an ecological consultancy providing a range of services to commercial and public sector clients (see www.biocensus.co.uk). The objective is to create an automatic system which can detect individual occurrences of the song of a particular species of bird over a period of several days, in order to establish whether or not that species is resident in the locality and, if so, to estimate the population size.
The project will involve applying pattern recognition techniques which have been demonstrated to be effective for computer speech recognition to birdsong. More specifically, the project will involve:

  • Selecting or developing microphone configurations suitable for capturing birdsong,
  • Collecting a database of birdsong 'in the field', focussing initial on the Black Redstart population in Birmingham
  • Understanding the structure of birdsong sufficiently well to propose signal processing methods which can extract features from birdsong which are suitable for species classification,
  • Applying techniques from automatic speech recognition, such as hidden Markov models (HMMs) and Gaussian Mixture Models (GMMs) to birdsong data
  • Applying noise-robustness techniques from computer speech recognition to the birdsong data, in order to make the classification process less sensitive to environmental noise
  • Off-line evaluation of the system using the database
  • Development of a prototype system and its evaluation in the field.

The project will be supervised jointly by Professor Martin Russell (www.eee.bham.ac.uk/russellm) and Dr Peter Jancovic (www.eee.bham.ac.uk/Jancovic). Biocensus will provide guidance on birdsong location, recognition and recording.
Funding sufficient to cover the cost of home fees plus £12,900 maintenance is available for a UK student only.
If you are interested please contact Martin Russell (m.j.russell@bham.ac.uk) or Peter Jancovic (p.jancovic@bham.ac.uk) for further information.