Audio Signal Processing and Coding

2.8 REVIEW OF MULTIRATE SIGNAL PROCESSING

Multirate signal processing (MSP) involves the change of the sampling rate while the signal is in the digital domain. Sampling rate changes have been popular in DSP and audio applications. Depending on the application, changes in the sampling rate may reduce algorithmic and hardware complexity or increase resolution in certain signal processing operations by introducing additional signal samples. Perhaps the most popular application of MSP is over-sampling analog-to-digital (A/D) and digital-to-analog (D/A) conversions. In over-sampling A/D, the signal is over-sampled thereby relaxing the anti-aliasing filter design requirements, and, hence, the hardware complexity. The additional time-resolution in the over-sampled signal allows a simple 1-bit delta modulation (DM) quantizer to deliver a digital signal with sufficient resolution even for high-fidelity audio applications. This reduction of analog hardware complexity comes at the expense of a data rate increase. Therefore, a down-sampling operation is subsequently performed using a DSP chip to reduce the data rate. This reduction in the sampling rate requires a high precision anti-aliasing digital low-pass filter along with some other correcting DSP algorithmic steps that are of appreciable complexity. Therefore, the over-sampling analog-to-digital (A/D) conversion, or as otherwise called Delta-Sigma A/D conversion, involves a process where complexity is transferred from the analog hardware ...