Abstract | ||
---|---|---|
We propose a training method for deep neural network (DNN) based source enhancement to increase objective sound quality assessment (OSQA) scores such as the perceptual evaluation of speech quality. In many conventional studies, DNNs have been used as a mapping function to estimate time-frequency masks and trained to minimize an analytically tractable objective function such as the mean squared err... |
Year | DOI | Venue |
---|---|---|
2018 | 10.1109/TASLP.2018.2842156 | IEEE/ACM Transactions on Audio, Speech, and Language Processing |
Keywords | DocType | Volume |
Training,Linear programming,Optimization,Speech processing,Quality assessment,Time-frequency analysis,Estimation | Journal | 26 |
Issue | ISSN | Citations |
10 | 2329-9290 | 7 |
PageRank | References | Authors |
0.52 | 18 | 5 |
Name | Order | Citations | PageRank |
---|---|---|---|
Koizumi Yuma | 1 | 41 | 11.75 |
Kenta Niwa | 2 | 95 | 17.07 |
Yusuke Hioka | 3 | 101 | 19.40 |
Kazunori Kobayashi | 4 | 12 | 1.76 |
Yoichi Haneda | 5 | 97 | 20.16 |