FIELD: speech analysis or synthesis; speech recognition.
SUBSTANCE: invention relates to signal processing means. Encoded low-frequency range signal and encoded energy information used to frequency shift the encoded low-frequency range signal are received. Encoded signal is decoded to obtain a decoded signal with an energy spectrum having a shape including an energy depression. Decoded signal is filtered, and by means of said filtering, the decoded signal is divided into low-frequency range band signals. Smoothing of the low-frequency range band signals is performed, and by means of said smoothing, the energy depression of the low-frequency range band signals is smoothed. Frequency shift of the smoothed low-frequency range band signals is performed, and by means of said frequency shift, high-frequency range band signals are generated from the low-frequency range band signals. High-frequency range band signals and low-frequency range band signals are combined to generate an output signal. Output signal is then output.
EFFECT: technical result is improved quality of the decoded signal.
3 cl, 14 dwg
