IP Library › Granted Patent US 11,581,000
Granted Patent B2
US 11,581,000 · App. 17/105,835 · Granted Feb 14, 2023

Apparatus and method for encoding/decoding audio signal using information of previous frame

Inventors: Woo-Taek Lim (Daejeon, KR); Seung Kwon Beack (Daejeon, KR); Jongmo Sung (Daejeon, KR); Mi Suk Lee (Daejeon, KR); Tae Jin Lee (Daejeon, KR)
Assignee: Electronics and Telecommunications Research Institute
G10L19/167G06N3/08G10L19/038G10L25/30
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 11,581,000
App. No.
17/105,835
Granted
Feb 14, 2023
Kind
B2
Abstract

Disclosed is an apparatus and method for encoding/decoding an audio signal using information of a previous frame. An audio signal encoding method includes: generating a current latent vector by reducing dimension of a current frame of an audio signal; generating a concatenation vector by concatenating a previous latent vector generated by reducing dimension of a previous frame of the audio signal with the current latent vector; and encoding and quantizing the concatenation vector.

Claims (8)

1. A method for encoding audio signal, comprising:

generating a current latent vector by reducing a dimension of a current frame of an audio signal;

generating a concatenation vector by concatenating a previous latent vector generated by reducing a dimension of a previous frame of the audio signal with the current latent vector; and

encoding and quantizing the concatenation vector to output a bit stream, and

wherein the generating the current latent vector reduces the dimension of the current frame of the audio signal using a neural network;

wherein the neural network is trained according to a loss function of the current latent vector calculated by setting the previous latent vector as a conditional probability.

2. The method of claim 1 ,

wherein the neural network is trained according to an entropy of the current latent vector calculated by setting the previous latent vector as a conditional probability.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Nov 27, 2020
From: LIM, WOO-TAEK; BEACK, SEUNG KWON; SUNG, JONGMO; LEE, MI SUK; LEE, TAE JIN
To: ELECTRONICS AND TELECOMMUNICATIONS RESEARCH INSTITUTE
Reel/Frame 055027/0263 →
Priority Claims (1)
KR 10-2019-0157168 · Nov 29, 2019 · national
Continuity (1)
Related Publication 20210166706A1 · Jun 3, 2021