PhD Supervisor: Paul White
The underwater environment is an extremely noise place (https://dosits.org/). The sources of these sounds are many-fold, including sounds from marine life (e.g. whales and dolphins), noise from physical processes (e.g. rain, waves and seismic activity) and from anthropogenic (man-made) sources. Passive sonar, in some contexts referred to as Passive Acoustic Monitoring, is the process of exploring and understanding the marine environment just by listening to the sounds heard. The passive nature of such systems mean that they are non-invasive so, for example, they do not disturb the natural behaviour on animals being studied. Such systems typically use an array of multiple hydrophones (underwater microphones) which allows them to gain information about which direction a sound is coming from.
The wide range of acoustic sources which can be observed, their time dependent nature and the variation with angle, all make the interpretation of underwater passive acoustic data a challenging task which has traditionally been undertaken manually. However, technical improvements mean that the number of hydrophones deployed at a time is rapidly increasing and systems are deployed for longer periods. All of which means that the amount of data collected is currently too large for manual processing and so computer based algorithms are of ever increasing importance, with the machine learning approaches offering the potential to dramatically improve our capability.
This project will focus on using machine learning approaches to enhance the ability to detect quiet sounds and so improve the sensitivity of passive system. It will explore the various approaches to novelty detection, seeking to identify new sound sources as they appear in the measurements. The challenge is that such an approach needs to be adaptive, since the background noise in the underwater world evolves over time, it must be able to adapt to the environment but remain computationally efficient, so that it can be used to process large data sets in realistic time-scales.
The project will be primarily computer based, it will focus on developing new methods based on pre-existing datasets.
This PhD is part sponsored by Thales and will be conducted with the Signal Processing and Hearing group (https://tinyurl.com/ahoupzkv) in the ISVR (https://tinyurl.com/auy7bqho). The sponsorship allows us to offer an enhanced stipend for the successful candidate, who will also be given the opportunity to spend 3 months with Thales during the project. The work will employ datasets from a range of sources, many of which already exist. This makes the PhD robust to conditions imposed by the pandemic, should they persist beyond the start of the project.
A very good undergraduate degree (at least a UK 2:1 honours degree, or its international equivalent).
Closing date: 14 May 2022
Funding: For UK students, Tuition Fees and a stipend of £20,285 tax-free per annum for up to 3.5 years.
How To Apply
Applications should be made online. Select programme type (Research), 2022/23, Faculty of Physical Sciences and Engineering, next page select “PhD Engineering & Environment (Full time)”. In Section 2 of the application form you should insert the name of the supervisor Paul White
Applications should include:
Two reference letters
Degree Transcripts to date
Apply online: https://www.southampton.ac.uk/courses/how-to-apply/postgraduate-applications.page
For further information please contact: email@example.com