Project 53: Automatic Speech Recognition of Accented Speech and Medical Terminology
Contact Information
Prof. Kai Yu
Email: kai.yu@sjtu.edu.cn
Project Description and Objectives
Automatic speech recognition is one of the fields that has seen great advances over the last few decades thanks to deep learning, however under non-cooperative conditions, the performance is still not ideal. Data imbalance and sparsity are difficulties in accented speech recognition with medical terminology. This leads to the key scientific problems of structured modeling of sparse data with uneven distribution. This project aims to investigate the implementation of transfer by learning to effectively improve the speech recognition effect of medical terms under different accents.
Eligibility Requirements
Interested students should have basic knowledge of machine learning.
Main Tasks
Development of a paradigm for accented speech recognition;
Completion of a research report.
Website