Using Advanced Deep Learning
School of Engineering Location: Bailrigg, Lancaster, UK Salary: £32,296 to £36,924 (Full time, fixed-term contract) Closing Date: Sunday 12 January 2025 Interview Date: To be confirmed Reference: 1336-24-R
Lancaster University's School of Engineering is seeking a highly motivated and talented Research Associate to join a cutting-edge project funded by the Defence and Security Accelerator Program. The project is in close collaboration with Collaboraite company as the end user and project partner. This exciting opportunity focuses on developing innovative Automatic Speech Separation and Enhancement Techniques using advanced deep learning approaches. The successful candidate will play a pivotal role at pushing the boundaries of using signal processing and data driven machine learning techniques for automatic speech recognition and enhancement tools.
About the Project
Automatic Speech Recognition (ASR) system is the engine used to accurately translate the spoken utterances into text and is an important component of the Automated Translation and Transcription tool proposed in this project. The performance of the ASR system can be degraded due to several sources of variability when using such a system in real-world scenarios. The first one is style of speech which can change from conversational or causal speech to formal and text-based speech and from continuous speech to isolated words pattern. Another source of variability is the difference in the speaker characteristics, i.e. the change in the speech rate, accent, etc. The last source of degradation is the environment where the speech is recorded and fed to the ASR system.
Although ASR systems have been dramatically improved in recent decades using data driven techniques, the speech recognition accuracy still significantly degrades in noisy environments. While many classical approaches have been developed to deal with this problem, they tend to be more effective in stationary noise such as white or pink noise than in the presence of more realistic degradations such as background music, background speech, and reverberation.
Robust ASR refers to the research field that addresses such performance degradation. Conventionally, the robustness of ASR models to background noise is improved by cascading speech separation and enhancement frontends and ASR backends. Speech separation refers to the case where the background is highly non-stationary and can contain difficult sources such as music or other speech signals. This problem has traditionally been addressed using model-based approaches, for example based on hidden Markov models (HMMs), or non- negative matrix factorization (NMF). More recently, however, various data- driven discriminative approaches, relying on deep learning techniques are proved to be effective for this task. Therefore, the objective of this project is to develop the machine learning and signal processing tools and techniques that can be used to improve the accuracy of monaural ASR systems in adverse real-world scenarios. To achieve this, we draw upon the recent theoretical advances in the field of data driven speech separation techniques in both time-domain and frequency-domain using deep neural networks.
Key Responsibilities:
Eligibility Criteria:
We strongly encourage candidates of all different backgrounds and identities to apply. To declare your interest and for further information, please send a copy of your CV along with the cover letter to Dr Allahyar Montazeri ([email protected]),
Apply Online
Further details:Please note: unless specified otherwise in the advert, all advertised roles are UK based.
Find out what it's like to work at Lancaster University, including information on our wide range of employee benefits, support networks and our policies and facilities for a family-friendly workplace.
The University recognises and celebrates good employment practice undertaken to address all inequality in higher education whilst promoting the importance and wellbeing for all our colleagues.
We warmly welcome applicants from all sections of the community regardless of their age, religion, gender identity or expression, race, disability or sexual orientation, and are committed to promoting diversity, and equality of opportunity.
Share: