Speech reconstruction for voice-loss patients
报告摘要: Many cases of voice loss in patients are due to problems with the vocal chords. This means that they can speak, but their voice is corrupted,sounding like hoarse whisper sounds. Most components of their speech still work properly, but the problem with their vocal chords means that they are not able to produce pitch, and their speech is almost totally unvoiced. In general, two methods of reconstructing their voice are statistical voice conversion (SVC) and parametric reconstruction. While SVC methods can have very high quality, they have some problems in practice. We therefore focus on parametric reconstruction, which is fast, efficient and very flexible. In this talk we will look briefly at SVC before considering some new advances in parametric reconstruction methods. 个人简介: Professor Ian McLoughlin is a Professor and Head of the School of Computing at the University of Kent Medway Campus in the UK. He was a professor at the University of Science and Technology of China NEL-SLIP lab from 2012-2015. Before that he spent 10 years at Nanyang Technological University, Singapore and 10 years in the electronics R&D industry in New Zealand and the UK. He obtained both his BEng and PhD from the University of Birmingham (UK). He became a Chartered Engineer in 1998 and a Fellow of the IET in 2013.He spent 15 years working for the electronics R&D industry and about the same length of time in academia. He has consulted and worked in 5 countries and presented his work to government science ministers of UK, Russia, USA, Malaysia, Taiwan, New Zealand and to the president of Singapore. He has over 200 papers, 4 books and 13 patents in the fields of speech & audio, wireless communications and embedded systems.