AUDIO DATA GENERATOR AND METHODS OF GENERATING AUDIO SIGNAL AND TRAINING AUDIO DATA GENERATOR Russian patent published in 2024 - IPC G10L13/02 

FIELD: physics.

SUBSTANCE: invention relates to computer engineering for processing audio data. Technical result is achieved by steps of: receiving, by a first processing unit, first data extracted from an input signal; for each channel of the first output data: processing target data by conditioning set of trained layers of the first processing unit to obtain parameters of conditioning features; and applying the conditioning feature parameters to the first data or to the normalized first data by means of the styling element of the first processing unit; outputting first output data comprising a plurality of channels by means of a first processing unit; receiving first output data or data extracted from first output data by means of second processing unit as second data; and the plurality of second data channels are combined by a second processing unit to obtain an audio signal.

EFFECT: high quality of generating an audio signal with limited computational resources.

54 cl, 12 dwg