Location: | Bath |
---|---|
Salary: | £29,605 to £36,024 Grade 6 |
Hours: | Full Time |
Contract Type: | Fixed-Term/Contract |
Placed On: | 5th July 2024 |
---|---|
Closes: | 16th July 2024 |
Job Ref: | ED11864 |
About the role
The Department of Computer Science seeks to recruit a Research Assistant in Speech and Natural Language Processing to work on reducing bias in automatic speech recognition systems.
This is a collaborative project between Wyser and the NLP Group at the University of Bath led by Dr Harish Tayyar Madabushi.
The project is funded by UKRI.
This is a fixed-term role, expected to end January 2025.
More than half of households in the UK with internet access use voice assistant devices like Amazon Alexa. These devices rely on ASR models, which use Natural Language Processing (NLP) at their core.
These models are increasingly used to help organisations interact with their customers and therefore the accuracy of those model is crucial for making decisions that can have far reaching impacts on people’s lives.
Linguistic bias occurs when the ASR model misinterprets what someone says due to factors like accent, dialect, or speech impairment.
By identifying and addressing these biases, this project will be supporting any organisation looking to use this form of AI to offer more accessible, fair and inclusive services.
The research assistant will be required to interface with the research and development team at Wyser, understand and document the different methods being experimented with and disseminate these findings by writing papers and, where possible presenting them at conferences.
As a member of Research Staff at the University of Bath, you will be encouraged to take up a minimum of 10 days professional development pro rata per year.
About You
The ideal candidate must have an undergraduate or, preferably, a master's degree in Computer Science or a closely related field with experience in Speech and Natural Language Processing.
Expertise in the use of Large Language Models is essential as is knowledge of pre-training, fine-tuning of language and speech models.
A significant aspect of the role involves understanding the specific methods being developed by the Wyser team.
The candidate must be capable of thoroughly documenting these methods to ensure they are recorded for future reference and publication.
The ideal candidate will have demonstrated experience in writing up experiments and research findings.
Given the requirement to work closely with the Wyser team, this role demands strong interpersonal skills and the ability to effectively communicate and integrate within a multidisciplinary team environment.
Type / Role:
Subject Area(s):
Location(s):