ABSTRACT
THE parallel implementation of a 5-band FIR-based equalizer was first discussed inChapter 3. A block diagram of such an equalizer and the generalized extension of this parallel implementation are shown in Figure 11.1. While this extension to M bands may not seem like a major change, increasing the computational complexity of the DSP algorithm by adding additional parallel filters will eventually result in being unable to meet the real-time schedule. At this stage in the equalizer’s development, we must either settle for the current level of system performance or rethink our approach to implementing the algorithm. This is very similar to the approach taken in Chapter 3 where we progressed from an easily understood brute force implementation of the filter’s dot-product, to a much more efficient, but more complicated to understand, implementation using a circular buffer.
Figure 11.1: Left: Block diagram associated with winDSK6’s 5-band Graphic Equalizer application. Right: generalized block diagram of an M -band graphic equalizer.
