Nobuaki Minematsu, University of Tokyo
http://www.gavo.t.u-tokyo.ac.jp/
Abstract: English is the only language available for global communication and is used by 1.5 billions of speakers. It is also known to have a large diversity of pronunciation due to the influence of speakers' mother tongue, called accents. Our project aims at creating a global and speaker-basis map of English accents to be used in learning World Englishes (WE) as well as research studies of WE. Creating the map, i.e., speaker-basis accent clustering, mathematically requires a distance matrix in terms of accents among all the speakers considered, and technically requires a method of predicting the accent distance between any pair of the speakers only by using their speech samples. In the talk, we present the results of our trials of prediction using structural analysis of speech and support vector regression. Two prediction experiments were done: speaker-pair-open and speaker-open experiments. A striking performance was obtained in the former but in the latter, good prediction was found to be difficult in the current framework.
Structural analysis of speech and its application to speaker-based phonetic clustering of World English pronunciations
Abstract: English is the only language available for global communication and is used by 1.5 billions of speakers. It is also known to have a large diversity of pronunciation due to the influence of speakers' mother tongue, called accents. Our project aims at creating a global and speaker-basis map of English accents to be used in learning World Englishes (WE) as well as research studies of WE. Creating the map, i.e., speaker-basis accent clustering, mathematically requires a distance matrix in terms of accents among all the speakers considered, and technically requires a method of predicting the accent distance between any pair of the speakers only by using their speech samples. In the talk, we present the results of our trials of prediction using structural analysis of speech and support vector regression. Two prediction experiments were done: speaker-pair-open and speaker-open experiments. A striking performance was obtained in the former but in the latter, good prediction was found to be difficult in the current framework.
0 comments:
Post a Comment