AUDIO ENCODER AND DECODER Russian patent published in 2016 - IPC G10L19/08 

FIELD: information technology.

SUBSTANCE: invention relates to multichannel audio encoding. Method of decoding in multichannel audio signal processing system for reconstruction of M encoded channels, where M>2 comprises following steps: reception of N down-mixing signals with coding shape, comprising spectral coefficients, corresponding to frequencies between first and second crossover frequencies, where 1<N<M; receiving M signals with coding shape, comprising spectral coefficients, corresponding to frequencies of up to first crossover frequency, wherein each of M signals with coding shape corresponds to corresponding one of M encoded channels; down-mixing M signals with coding shape in N down-mixing signals, containing spectral coefficients which correspond to frequencies of up to first crossover frequency; combination of each of N down-mixing signals with coding shape, comprising spectral coefficients, corresponding to frequencies between first and second crossover frequencies with corresponding one of N down-mixing signals, comprising spectral coefficients, corresponding to frequencies of up to first crossover frequency, in N combined down-mixing signals; expansion of each of N combined down-mixing signal into a frequency range above second crossover frequency by means of high frequency reconstruction, as a result of which each expanded down-mixing signal comprises spectral coefficients, corresponding to a range passing below first crossover frequency and higher than second crossover frequency; performing parametric up-mixing of N combined down-mixing signals with extended frequency range in M up-mixing signals, comprising spectral coefficients, corresponding to frequencies higher than first crossover frequency, each of M up-mixing signals corresponds to one of M encoded channels; and combination of M up-mixing signals comprising spectral coefficients, corresponding to frequencies higher than first crossover frequency, with M signals with coded shape, comprising spectral coefficients which correspond to frequencies of up to first crossover frequency.

EFFECT: high quality of encoded and decoded audio signal.

29 cl, 8 dwg