Methods for Encoding and Decoding Audio Signal Using Neural Network Model, and Devices for Performing the Same

摘要

Disclosed are a method for encoding and decoding an audio signal using a neural network model, a training method for a neural network model, and an encoder and decoder for performing the same. An encoding method according to an embodiment of the present invention comprises the steps of: generating first characteristic information for the input signal using a cyclic encoding model; generating an output signal from the first feature information using a cyclic decoding model; determining a residual signal based on a difference between the input signal and the output signal; generating second characteristic information about the residual signal by using an acyclic coding model; and converting the first characteristic information and the second characteristic information into a bitstream.