Environment-aware Listener-optimised Binaural Enhancement of Speech
Research type
Research Study
Full title
Environment-aware Listener-optimised Binaural Enhancement of Speech
IRAS ID
195015
Contact name
Stuart Rosen
Contact email
Sponsor organisation
University College London
Clinicaltrials.gov Identifier
Z6364106/2015/11/34 , Data Protection Registration
Duration of Study in the UK
1 years, 5 months, 31 days
Research summary
Hearing loss affects over half the UK population aged over 60. Even with hearing aids, understanding speech in noisy environments is particularly challenging for hearing impaired listeners. One approach to improving speech perception in noise via hearing aids involves the application of computerised signal processing methods. For example, a technique known as binary masking is intended to effectively reduce the amount of noise by identifying and preserving portions of the mixed speech and noise signal that contain relatively high amounts of speech, while eliminating parts of the mixture that are mostly noise. Speech enhancement based on binary masking has been shown to make noisy speech more intelligible for both normal and hearing-impaired listeners, in some conditions.
However, a shortcoming of existing methods is that they do not preserve the advantages derived from having two ears rather than one. Sounds that originate from off to one side arrive at one ear slightly before the other and are slightly louder in that ear due to sound energy being absorbed as it travels through the head. The auditory system uses these differences to separate out sounds coming from different directions, which can be very beneficial when listening to speech in a noisy environment. A major goal of this project is to extend existing binary masking techniques to provide speech enhancement that preserves these differences between ears.
The optimal form of enhancement depends upon both the listening environment and the listener’s particular pattern of hearing loss. The techniques developed will take into account information on the listener’s hearing abilities, and in addition, will assess the acoustic environment, recognise typical listening situations and automatically adjust the enhancement applied.
The adequacy of the techniques developed will be assessed for both speech intelligibility in noise and sound quality by test conducted in a group of hearing-impaired listeners typical of those seen clinically.REC name
London - Brighton & Sussex Research Ethics Committee
REC reference
17/LO/0151
Date of REC Opinion
11 Apr 2017
REC opinion
Further Information Favourable Opinion