Development of a Speaker Verification Method Based on Phase Corrected by Vocal Source Information - Toward Forensic Science Application

2022 Fiscal Year Final Research Report

• Project/Area Number: 20K23357
• Research Category: Grant-in-Aid for Research Activity Start-up
• Allocation Type: Multi-year Fund
• Review Section: 1002:Human informatics, applied informatics and related fields
• Research Institution: National Research Institute of Police Science
• Principal Investigator: Okada Masahiro 科学警察研究所, 法科学第四部, 研究員 (80874502)
• Project Period (FY): 2020-09-11 – 2023-03-31
• Keywords: 位相補正 / Electroglottography / 声道フィルタの位相特性 / 話者性・個人性 / 発話スタイル / 角度統計 / 平均位相スペクトル

Outline of Final Research Achievements

In this study, we developed a speaker estimation method based on phase corrected with the combination of speech and glottal sources. For speaker verification by visual observation, the phase characteristics of the vocal tract filter were visualized from real data of speech and glottal sources. As a result, the visualized phase characteristics were consistent for the same speaker in the same-sentence utterance and were different for different speakers despite the same-sentence utterance. Furthermore, for quantitative evaluation, we devised an average phase spectra that aggregated the visualized phase information and conducted a speaker verification experiment using the spectra. The experimental results showed that the accuracy for speaker verification was high regardless of the verification threshold. Therefore, it was clarified that speaker verification based on phase information was effective.

Free Research Field

音声情報処理

Academic Significance and Societal Importance of the Research Achievements

特殊詐欺などの音声が関連する犯罪では証拠として残る録音は短く、犯人と思しき話者を推定するには手がかりが少ない。そのような情報が乏しい状況にもかかわらず、通常の分析法では音声に含まれる振幅のみが使用され、位相は用いられない。本研究は位相の話者識別能力を検証するものであり、それが示された今、位相利用が促進され、犯罪捜査へ貢献できるものと考えられる。また、学術的には、これまで着目されてこなかった位相への注目度が高まることが期待される。