Title:
VOICE SYNTHESIS APPARATUS USING ARTIFICIAL INTELLIGENCE, OPERATING METHOD FOR VOICE SYNTHESIS APPARATUS, AND COMPUTER-READABLE RECORDING MEDIUM
Document Type and Number:
WIPO Patent Application WO/2020/166748
Kind Code:
A1
Abstract:
A voice synthesis apparatus according to one embodiment of the present invention comprises: a memory for storing a plurality of sentences and, for each sentence, preliminary information about a word classified as a minor class from among a plurality of classes; and a processor for determining the oversampling rate of the word on the basis of the preliminary information, determining the number of times of oversampling of the word by using the determined oversampling rate, and generating, as many as the determined number of times of oversampling, sentences including the word, wherein the plurality of classes includes a first class corresponding to a first degree of spaced reading, a second class corresponding to a second degree of spaced reading that is greater than the first degree of spaced reading, and a third class corresponding to a third degree of spaced reading that is greater than the second degree of spaced reading, and the minor class can be the class with the smallest number in one sentence from among the first to third classes.
Inventors:
CHAE JONGHOON (KR)
HAN SUNGMIN (KR)
HAN SUNGMIN (KR)
Application Number:
PCT/KR2019/001886
Publication Date:
August 20, 2020
Filing Date:
February 15, 2019
Export Citation:
Assignee:
LG ELECTRONICS INC (KR)
International Classes:
G10L13/08; G10L13/02
Foreign References:
| KR20190002812A | 2019-01-09 | |||
| KR20170011636A | 2017-02-02 | |||
| KR101179915B1 | 2012-09-06 | |||
| KR20150066361A | 2015-06-16 | |||
| US8356032B2 | 2013-01-15 |
Attorney, Agent or Firm:
HAW, Yong Noke (KR)
Download PDF: