|Funding for:||UK Students, EU Students, International Students|
|Funding amount:||University funded|
|Placed On:||4th August 2022|
|Closes:||18th September 2022|
Speech impairment (through a range of conditions that cause speech dysfunctionality) causes considerable anxiety and distress for people, and importantly in many cases considerably restricts their ability to verbally communicate. The issue is widespread, and in the UK alone there are over 14 million people with such speech dysfunctionality conditions that are classed as speech impairments. In addition, speech impairment also occurs through clinical conditions such as cancers affecting the head and neck.
An EPSRC-Loughborough funded study over the past 4 years has explored a new concept to help interpret communication through breath-activated Augmentative and Alternative Communication (AAC) technology, in which a prototype AAC device comprising a pressure sensor and associated electronics/hardware and software (including fully developed app) has gained proof of concept for interpreting and translating pressure and sound variations from breath analysis on 48 healthy volunteers at Loughborough University.
Currently, this system relies on pattern recognition through dynamic time warping and a k-NN classification technique, which although effective, is computationally intensive for high sample rates limiting its scalability and speed. This project aims to improve on these methods by considering end-to-end techniques based on modern machine learning, exploiting modern neural network topologies such as LSTMs and/or convolutional networks to provide an accurate, computationally efficient AAC solution that may be deployed to appropriate hardware through code generation from MATLAB/Simulink, to test the feasibility of such methods on both smartphone and embedded hardware.
Is the project University funded or self-funded? University funded
Funding eligibility: Competition funded project (students worldwide)
Who is eligible to apply? Both UK and International
Full-time/part-time availability: Full-time (3 years)
Closing Date: 18th September 2022
Advert Reference: P2SAM22-08
Type / Role: