公开(公告)号：US20230326445A1

公开(公告)日：2023-10-12

申请号：US17658807

申请日：2022-04-11

Applicant: Snap Inc.

Inventor： Guy Adam ,  Jackie Assa ,  Alan Bekker

IPC: G10L13/08 ,  G06N20/00 ,  G10L15/187 ,  G10L15/06 ,  G06T13/20 ,  G06T13/40

CPC classification number: G10L13/08 ,  G06N20/00 ,  G10L15/187 ,  G10L15/063 ,  G06T13/205 ,  G06T13/40

Abstract: Systems and methods are provided for providing animated speech refinement. The systems and methods perform operations comprising: receiving an audio stream comprising one or more spoken words; processing the audio stream by an automated speech recognition (ASR) engine to identify base timing of one or more phonemes corresponding to the one or more spoken words; applying a machine learning model to the base of the one or more phonemes to estimate an adjustment to the base timing of the one or more phonemes.