| 摘要 |
PURPOSE: A method for quantizing vectors of line spectrum for encoding aural signals and a method for calculating a masking threshold therefor are provided to improve auditory distortion generated in generating aural signals, and realize LSF vector quantization by using fewer bits. CONSTITUTION: An aural sample code to be compared with an input LSF(Line Spectral Frequency) is extracted from a code book. A predetermined auditory weighted euclidean distance is applied to calculate an error distance according to auditory distortion. The auditory weighted euclidean distance considers a first weight according to a weighted euclidean distance between LSF coefficients of the extracted aural sample code and the input LSF vector, and a second weight according to a masking effect. An aural sample code minimizes the error distance according to auditory distortion is output as a quantized code ward. |
