Engineering and Materials and Computing and Informatics Research Seminar

omputing Seminar: The impact of automatic exaggeration of the visual articulatory features of a talker on the intelligibility of spectrally distorted speech

Light bulb icon
Seminars

As part of the School of Science and Technology's Joint Seminar Programme for the Engineering and Materials and Computing Informatics Research Centres, Dr Najwa Alghamidi, University of Sheffield presents: Computing Seminar: The impact of automatic exaggeration of the visual articulatory features of a talker on the intelligibility of spectrally distorted speech

  • From: Wednesday 31 January 2018, 1 pm
  • To: Wednesday 31 January 2018, 2 pm
  • Location: 015, CELS, Nottingham Trent University, Clifton Campus, Clifton Lane, Nottingham, NG11 8NS

Past event

Event details

As part of the School of Science and Technology's Joint Seminar Programme for the Engineering and Materials and Computing Informatics Research Centres, Dr Najwa Alghamidi, University of Sheffield presents: Computing Seminar: Computing Seminar: The impact of automatic exaggeration of the visual articulatory features of a talker on the intelligibility of spectrally distorted speech

Abstract:

The robustness of human speech perception arises from a listener’s ability to integrate and evaluate information from multiple sources. Audiovisual integration refers to a listener’s ability to utilise heard (auditory) and seen (visual) speech information in order to interpret the perceived message from the talker. In daily life, we tend to vary our mode of speaking from hypo-articulation – a lazy way of speaking – to hyper-articulation, in which we enhance our speech by exerting more articulation energy and vocal effort to improve the perceptual quality of communications. Such habits have inspired researchers to develop speech enhancement methods that aim to improve the intelligibility, or the clarity, of the heard speech. Despite the importance of the visual part of speech, which can improve speech intelligibility by up to 22 dB, the field of visual speech enhancement has attracted little interest to date. Such enhancement would benefit those who have limited access to heard speech and who show increased bias to visual speech, such as the hard of hearing.

This talk takes a look at possible methods for visual speech enhancement as well as these methods’ applications. The talk will also explore an example of natural enhancement in visual speech by investigating the visual modifications observed in speech produced within what is known as the Lombard effect, which is an unconscious reaction that is regulated by self-monitoring of the voice and by the need to maintain intelligible communication during noisy conditions.

This seminar is hosted by Dr Richard Cant.

Location details

Room/Building:

015, CELS

Address:

Nottingham Trent University
Clifton Campus
Clifton Lane
Nottingham
NG11 8NS

Past event

Still need help?

+44 (0)115 941 8418