Audio coding method based on spectral recovery scheme

A decoding method of a frequency spectrum recovery scheme performed by a decoder, the method comprising:

• decoding a first frequency spectral coefficient transmitted from an encoder;
• recovering a magnitude of a second frequency spectral coefficient through a neural network;
• to determining a sign transmission rule of a second frequency spectral coefficient, which is not transmitted by the encoder;
• allocating a sign of the second frequency spectral coefficient depending on the sign transmission rule;
• and
• performing inverse transform coding on the first frequency spectral coefficient and a second frequency spectrum, of which the sign and the magnitude are determined, to output a recovered input signal.

A decoder performing a decoding method of a frequency spectrum recovery scheme, the decoder comprising:

• a processor,
• wherein the processor is configured to:
• decode a first frequency spectral coefficient transmitted from an encoder;
• recover a magnitude of a second frequency spectral coefficient through a neural network;
• determine a sign transmission rule of a second frequency spectral coefficient, which is not transmitted by the encoder;
• allocate a sign of the second frequency spectral coefficient depending on the sign transmission rule;
• and
• perform inverse transform coding on the first frequency spectral coefficient and a second frequency spectrum, of which the sign and the magnitude are determined, to output a recovered input signal.