| 摘要 |
PURPOSE: To provide a method and device by which subbands can be filtrated at a high speed in encoding/decoding MPEG speech signals. CONSTITUTION: An input sample is mapped to a frequency coefficient by using a subband analyzing filter (102) and a scaling factor is transmitted at every continuous output sample of each subband (103). In parallel with the transmission, an input speech signal is Fourier-transformed at a high speed (101) and a masking threshold is decided (110). Then a timbre signal component, non- timbre signal, and silence signal are calculated (111) and bit assignment is repeatedly calculated (112). Other bits are calculated for encoding each frame (106) and bit assignment information is transmitted (107). In addition, an encoded bit stream is outputted from an encoder for storage and arithmetic processing of addition, etc., (108). |
