Scalable and embedded codec for speech and audio signals

A system and method for processing audio or speech signals into digital code which is scalable to different digital sampling rates or transmitted bit rates. The input signal spectrum is split by band splitter (5) into portions based on selected frequencies, Bi, as shown on Fig. 1B. Each portion of the split spectrum is separately inputted to embedded encoders (2) to be packetized for particular applications by packet assembler (7) as shown in the general case of embedded coding by Fig. 1A; for windowed FFT octave bands by Fig. 2A; and for the corresponding inverse process at the receiver end by Fig. 2B. Fig. 2B shows packet extraction of coded signal by (9) and inputting of extracted code into corresponding embedded decoders (4) wherein the separate windows required for an application are assembled by (3) to produce an output signal at the desired sampling rate. A sinusoidal transform coder is shown by Fig. 3A where embeddedness is performed by reconstructions based on selected parameters such as pitch, voicing probability, sinusoid amplitudes and phase measurements for higher-bit-rate applications.