FIELD: computer engineering.
SUBSTANCE: processing audio data in a way increasing the accuracy of the original content of the audio signal, eliminating the introduction of distortions in the form of audio signal packaging artefacts. It is achieved by receiving an audio signal containing a number of time segments, determining the appropriate gain for each time segment of the audio signal, while the gain is in the frequency domain and is based on the p-norm of the spectral amplitudes of each time segment of the representation in the frequency domain of the audio signal, and this value of the p-norm is chosen so as to highlight the spectral composition of the weak signal of the audio signal relative to the spectral composition of the strong signal of the audio signal; and applying appropriate gain values to each time segment to obtain a decompressed audio signal, wherein applying the respective individual gain values boosts high intensity segments and attenuates low intensity segments, and each corresponding gain for each time segment is computed using subband samples in a subset of subbands in the corresponding time segment.
EFFECT: increasing the accuracy of the original content of the audio signal, eliminating the introduction of distortions in the form of audio signal packaging artefacts.
8 cl, 8 dwg
