Title:
SPEAKER DIARIZATION METHOD, SPEAKER DIARIZATION DEVICE, AND SPEAKER DIARIZATION PROGRAM
Document Type and Number:
WIPO Patent Application WO/2022/130471
Kind Code:
A1
Abstract:
An array generation unit (15b) splits a per-frame acoustic feature sequence of an acoustic signal into segments of a predetermined length and generates an array in which a plurality of split segments in the row direction are disposed in the column direction. A training unit (15d) uses the generated array to generate, by training, a speaker diarization model (14a) that infers a speaker label of a speaker vector of each frame.
Inventors:
ANDO ATSUSHI (JP)
MURATA YUMIKO (JP)
MORI TAKESHI (JP)
MURATA YUMIKO (JP)
MORI TAKESHI (JP)
Application Number:
PCT/JP2020/046585
Publication Date:
June 23, 2022
Filing Date:
December 14, 2020
Export Citation:
Assignee:
NIPPON TELEGRAPH & TELEPHONE (JP)
International Classes:
G10L17/00; G10L15/04; G10L17/02; G10L17/04; G10L17/18
Foreign References:
| JP2019086679A | 2019-06-06 |
Attorney, Agent or Firm:
SAKAI INTERNATIONAL PATENT OFFICE (JP)
Download PDF: