Development of high-level speech processing infrastructure based on static representations of auditory information

2017 Fiscal Year Final Research Report

• Project/Area Number: 15H02726
• Research Category: Grant-in-Aid for Scientific Research (B)
• Allocation Type: Single-year Grants
• Section: 一般
• Research Field: Perceptual information processing
• Research Institution: Wakayama University
• Principal Investigator: Kawahara Hideki 和歌山大学, 学内共同利用施設等, 名誉教授 (40294300)
• Co-Investigator(Kenkyū-buntansha): 西村 竜一 和歌山大学, システム工学部, 助教 (00379611) ; 松井 淑恵 豊橋技術科学大学, 工学研究科, 准教授 (10510034) ; 坂野 秀樹 名城大学, 理工学部, 准教授 (20335003) ; 入野 俊夫 和歌山大学, システム工学部, 教授 (20346331) ; 森勢 将雅 山梨大学, 大学院総合研究部, 准教授 (60510013) ; 榊原 健一 北海道医療大学, リハビリテーション科学部, 准教授 (80396168) ; 戸田 智基 名古屋大学, 情報基盤センター, 教授 (90403328)
• Research Collaborator: Schweinberger Stefan Jena大学(独)
• Project Period (FY): 2015-04-01 – 2018-03-31
• Keywords: 音声情報処理 / 音声分析変換合成 / 基本周波数 / 声帯音源 / 情報表現 / 深層学習 / オープンソース

Outline of Final Research Achievements

We developed infrastructures of speech analysis, modification, and synthesis based on interference-free representations of speech parametric representations. In addition to our STRAIGHT-based infrastructure, which is a defacto standard in speech research, we developed a set of new independent algorithms. We made these algorithms as open-source. We elaborated on building supporting tools for promoting academic research using STRAIGHT systems. In addition to these planned accomplishments, we also established application infrastructure based on WaveNet, which revolutionalized speech applications based on deep learning.

Free Research Field

聴覚メディア処理