Digital Signal Processor: Sub-band Coding

Digital encoding of speech and audio has been a topic of longstanding interest for purposes of digital communications and digital storage1"3. The efficiency of such encoding techniques depends strongly on the degree to which the bit rate can be reduced (compressed) without impairing the quality of the decoded signal. Typically, signals such as speech and audio have a high degree of redundancy that can be used to reduce this bit rate. Also, properties of human perception can be used to reduce the bit rate without impairing the quality of the decoded signal. To take advantage of these properties, a considerable amount of signal processing is necessary. Thus, in the past many of these techniques have only been implemented by non-real-time computer simulations or with the aid of highly specialized digital hardware. This 1633 picture is now rapidly changing, as is exemplified by the recent Bell Laboratories digital signal processing integrated circuit (DSP).4,5 With this device it is possible to conveniently implement, in real time, signal processing algorithms of low to medium complexity. Thus, a single DSP integrated circuit can be used to implement many of the simpler encoding algorithms and multiple DSPS can be used for some of the more complex algorithms. In a companion paper, 6 it is shown that the ADPCM (adaptive differential PCM) encoding algorithm, which offers a bit rate reduction factor of approximately two over conventional logarithmic companded PCM encoding (for speech), can be efficiently implemented on the DSP, and that it uses only about one-quarter of the processing capability of the device.