2023 Fiscal Year Final Research Report
Transfer characteristics of emotional speech information toward elderly persons with hearing loss and development of novel speech morphing methods
| Project/Area Number |
21K19794
|
|---|
| Research Category |
Grant-in-Aid for Challenging Research (Exploratory)
|
| Allocation Type | Multi-year Fund |
|---|
| Review Section |
Medium-sized Section 61:Human informatics and related fields
|
|---|
| Research Institution | Wakayama University |
|---|
Principal Investigator |
Irino Toshio 和歌山大学, システム工学部, 教授 (20346331)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
松井 淑恵 豊橋技術科学大学, 次世代半導体・センサ科学研究所, 教授 (10510034)
森勢 将雅 明治大学, 総合数理学部, 専任准教授 (60510013)
|
|---|
| Project Period (FY) |
2021-07-09 – 2024-03-31
|
| Keywords | 高齢難聴 / 模擬難聴 / 聴覚情報処理 / 聴覚モデル / 感情知覚 / 弁別実験 / 音声分析合成 / 音声モーフィング |
|---|
| Outline of Final Research Achievements |
Using a voice morphing tool and a hearing loss (HL) simulator, emotion discrimination experiments were conducted on combinations of three emotions, "anger," "sadness," and "happiness," for young normal hearing (YNH) participants listening to normal and simulated HL sounds. The results showed that age-related HL in the peripheral auditory system does not affect emotion perception. We also conducted experiments with older participants and found that they had more difficulty discriminating between "anger" and "sadness" than other pairs for which discrimination performance was similar to that of YNH. In addition, a prototype of an automatic voice morphing tool was developed to facilitate future experiments. It was shown that morphing with better sound quality than conventional methods is possible without using the previously required phoneme sequence information. We believe that this is an important step towards full automation in the future.
|
| Free Research Field |
聴覚計算理論/心理実験/音信号処理
|
| Academic Significance and Societal Importance of the Research Achievements |
高齢者への音声伝達支援のために、補聴器が従来から用いられてきた。しかし、言語内容の伝達には役立つが、感情伝達の助けとはならない。一方、感情知覚研究に関しては、心理的な知見だけで、聴覚モデル化は意識されてこなかった。本研究により、聴覚モデル化も視野に入れた基礎的な実験な知見を得ることができた。これによりお年寄りに気持ちが十分伝わる新しい補聴アルゴリズムの可能性が出てきた。さらに、音声モーフィング完全自動化に向けた大きな前進もあった。これは、単に実験刺激作成にとどまらず、さまざまな音声合成技術に適用可能で、たとえばロボットと人間とのより良いインタラクションにも活用できると考えられる。
|
