Chi tiết| Tên | Improving Acoustic Model for Vietnamese Large Vocabulary Continuous Speech Recognition System using Deep Bottleneck features |
| Lĩnh vực | Tin học |
| Tác giả | Quoc Bao Nguyen, Tat Thang Vu, Chi Mai Luong |
| Nhà xuất bản / Tạp chí | |
| Số hiệu ISSN/ISBN | |
| Tóm tắt nội dung | |
In this paper, a method based on deep learning for extractingbottleneck features for Vietnamese large vocabulary speech recognitionis presented. Deep bottleneck features (DBNFs) is able to achieve signi cant improvements over a number of base bottleneck features whichwas reported previously. The experiments are carried out on the datasetcontaining speeches on Voice of Vietnam channel (VOV). The resultsshow that adding tonal feature as input feature of the network reachedaround 20% relative recognition performance. The DBNF extraction forVietnamese recognition decrease the error rate by 51%, compared to theMFCC baseline.
|
|
