Title:
VOICE FILE FORGERY DETECTION DEVICE AND METHOD THEREFOR
Document Type and Number:
WIPO Patent Application WO/2024/090608
Kind Code:
A1
Abstract:
The present invention relates to a voice file forgery detection apparatus using a specific pronunciation and a method therefor. According to the present invention, a voice file forgery detection device using a specific pronunciation comprises: a receiving part that receives a recorded voice file through a user terminal; a preprocessing part that outputs the received voice file in the form of a graph consisting of a time axis and a frequency axis, and independently amplifies a specific transition band in the outputted graph; a signal detection part that extracts a specific signal according to an air ejection phenomenon from the specific transition band, and extracts a voice waveform corresponding to an aspirated sound or an alveolar-palatal fricative; a forgery determination part that analyzes the correlation between the specific signal and the voice waveform, and determines whether the voice file has been forged according to the analyzed result; and a control part that displays and outputs the forged section on a screen when the voice file is determined to have been forged. According to the present invention as above, the transition band is independently amplified from the voice file outputted in the form of a graph, and the air ejection phenomenon generated by the amplified transition band is used to determine whether the voice file is forged, and thus it is possible to detect forgery even in voice files that have undergone mix-pasting and compression, and as it is possible to objectively detect the edited section of the voice file, the voice file can be used as objective evidence.
Inventors:
PARK JAE WAN (KR)
Application Number:
PCT/KR2022/016527
Publication Date:
May 02, 2024
Filing Date:
October 27, 2022
Export Citation:
Assignee:
SSMM INC (KR)
FOUNDATION SOONGSIL UNIV INDUSTRY COOPERATION (KR)
FOUNDATION SOONGSIL UNIV INDUSTRY COOPERATION (KR)
International Classes:
G10L25/15; G10L25/48
Foreign References:
| KR101566425B1 | 2015-11-06 | |||
| KR20210079060A | 2021-06-29 | |||
| KR20080017511A | 2008-02-27 | |||
| KR20220144997A | 2022-10-28 |
Other References:
JEONG SANG HUN: "The Study on Phonetical Information for Speech Recognition.", THE RESEARCH ON KOREAN LANGUAGE AND LITERATURE, THE SOCIETY OF KOREAN LITERARY STUDIES (OLD DONG-AK SOCIETY OF LANGUAGE AND LITERATURE, vol. 49, 1 August 2007 (2007-08-01), pages 135 - 160, XP093163286
NAM IN PARK: "A Study on Authentication Analysis Procedure of Digital Audio Files", JOURNAL OF DIGITAL FORENSIC, 한국디지털포렌식학회, vol. 13, no. 4, 31 December 2019 (2019-12-31), pages 257 - 270, XP093163297, ISSN: 1976-5304, DOI: 10.22798/kdfs.2019.13.4.257
NAM IN PARK: "A Study on Authentication Analysis Procedure of Digital Audio Files", JOURNAL OF DIGITAL FORENSIC, 한국디지털포렌식학회, vol. 13, no. 4, 31 December 2019 (2019-12-31), pages 257 - 270, XP093163297, ISSN: 1976-5304, DOI: 10.22798/kdfs.2019.13.4.257
Attorney, Agent or Firm:
YOUNGBEE INTELLECTUAL PROPERTY (KR)
Download PDF: