DEVICE AND METHOD FOR PRIVACY-PRESERVING VOCAL INTERACTION

The present invention relates to a home supporting device (100′) and method for privacy-preserving vocal interaction. A microphone (102) captures an audio signal (102) corresponding to a voice user query. The identity (122) of a speaker is determined and an obfuscated name (126) is generated corresp...

Full description

Saved in:
Bibliographic Details
Main Authors NEUMANN CHRISTOPH, FONTAINE PATRICK
Format Patent
LanguageEnglish
Korean
Published 17.06.2019
Subjects
ACOUSTICS
CALCULATING
COMPUTING
COUNTING
ELECTRIC DIGITAL DATA PROCESSING
MUSICAL INSTRUMENTS
PHYSICS
SPEECH ANALYSIS OR SYNTHESIS
SPEECH OR AUDIO CODING OR DECODING
SPEECH OR VOICE PROCESSING
SPEECH RECOGNITION
Online AccessGet full text

Cover

More Information
Summary:The present invention relates to a home supporting device (100′) and method for privacy-preserving vocal interaction. A microphone (102) captures an audio signal (102) corresponding to a voice user query. The identity (122) of a speaker is determined and an obfuscated name (126) is generated corresponding to the identified speaker. The audio signal (102) is analyzed to determine the intention (123) of a user, and a personalized response (124) is generated in combination with the obfuscated name. Then, the response is de-obfuscated by reintroducing the name of the speaker. The de-obfuscated response (127) is rendered to the speaker. 프라이버시-보존 음성 상호작용을 위한 홈 지원 디바이스(100') 및 방법. 마이크로폰(102)은 음성 사용자 질의에 대응하는 오디오 신호(102)를 캡처한다. 스피커의 아이덴티티(122)가 결정되고, 난독화된 이름(126)이 식별된 스피커에 대응하여 생성된다. 오디오 신호는 사용자의 의도(123)를 결정하도록 분석되고, 개인화된 응답(124)은 난독화된 이름과 결합하여 생성된다. 그 후 이러한 응답은 스피커 이름을 재도입함으로써 탈-난독화된다. 탈-난독화된 응답(127)은 스피커에 랜더링된다.
Bibliography:Application Number: KR20180154627