公开(公告)号：US20220310077A1

公开(公告)日：2022-09-29

申请号：US17592956

申请日：2022-02-04

Applicant: SAMSUNG ELECTRONICS CO., LTD.

Inventor： Mei TU ,  Fan ZHANG ,  Wangli CHAO ,  Xiaoyun XU ,  Song LIU ,  Shuo HU ,  Xue WEN ,  Liming SONG ,  Xiaoyan LOU

IPC: G10L15/187 ,  G10L15/22 ,  G10L15/00 ,  G10L15/02

Abstract: A speech recognition method, an apparatus, an electronic device, and a computer-readable storage medium are provided. The method includes acquiring a first speech recognition result of a speech; acquiring context information and pronunciation feature information about a target text unit in the first speech recognition result; and acquiring a second speech recognition result of the speech based on the context information and the pronunciation feature information.

公开(公告)号：US20240046920A1

公开(公告)日：2024-02-08

申请号：US18473701

申请日：2023-09-25

Applicant: Samsung Electronics Co., Ltd.

Inventor： Xiaoyan LOU ,  Fan ZHANG ,  Xiangfeng MENG ,  Liming SONG

IPC: G10L15/08 ,  G10L25/51

CPC classification number: G10L15/08 ,  G10L25/51 ,  G10L2015/088

Abstract: A data processing method, a device wake-up method, an electronic device, and a storage medium are provided. In the data processing method, speech to be processed is converted into a keyword phone sequence, and a similar pronunciation sequence generator acquires a similar phone sequence corresponding to the keyword phone sequence in a sequence generation manner, thereby acquiring a first data processed result corresponding to the speech to be processed. By replacing the search method of large-scale speech databases with this generation manner, effective coverage of possible real-life sounds can be achieved with a smaller model, thus improving the ability to distinguish confusing pronunciations. The above data processing method performed by the electronic device can be performed by an artificial intelligence (AI) model.