2023 Fiscal Year Final Research Report
Speech security on human-computer interaction
| Project/Area Number |
22K21304
|
|---|
| 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 | Japan Advanced Institute of Science and Technology |
|---|
Principal Investigator |
MAWALIM CandyOlivia 北陸先端科学技術大学院大学, 先端科学技術研究科, 助教 (10963720)
|
|---|
| Project Period (FY) |
2022-08-31 – 2024-03-31
|
| Keywords | voice privacy / phase vocoder / speaker anonymization / speaker verification / spoof attacks / speech intelligibility / auditory model |
|---|
| Outline of Final Research Achievements |
In FY2022, we developed speaker anonymization methods using time-scale modification. The phase vocoder method is most effective for preserving voice characteristics. This method offered a better balance between privacy and speech intelligibility. Additionally, we analyzed the impact of anonymization on gender perception using a machine learning model. These findings were presented at three international conferences.
In FY2023, research focused on two areas: (1) addressing unclear goals in speaker anonymization and the variety of speech attacks. New methods for tackling spoofing in speaker verification systems were developed. These findings were presented at two conferences. (2) investigating human speech perception to understand how we perceive intelligibility. This research, published in the Journal of Applied Acoustics, lays the groundwork for detecting changes caused by speech synthesis. Finally, the project is expanding its scope to include developing a Thai language spoof database.
|
| Free Research Field |
speech security, voice privacy
|
| Academic Significance and Societal Importance of the Research Achievements |
Innovative techniques for speaker anonymization and spoofing detection open up new possibilities for voice privacy and security research. This research will greatly contributes to securing voice communication, strengthening authentication systems, and improving human-computer interaction.
|
