The starting point of this chapter was the computation of a time-frequency representation of a sound, to manipulate this representation and reproduce a sound. At first sight this may appear as an easy task, but we have seen that the basis for this time-frequency processing needs a careful description of the fundamentals, because the term ‘vocoder’ can cover different implementations. We also explained that the arbitrary manipulation of time-frequency representations renders images in a way that they are no longer time-frequency representations of “real” sounds. This phenomenon leads to artifacts, which cannot be avoided.
Digital audio effects described in this chapter only perform manipulations of these time-frequency representations. These effects exclude the extraction of resonances, which will be the subject of the next chapter, and high-level processing such as the extraction of sinusoids and noise. For example, the mentioned bank of filters does not assume any parametric model of the sound. Nevertheless such effects are numerous and diverse. Some of them have brought new solutions to well-known techniques such as filtering. Pitch shifting and time stretching have shown their central place in the phase vocoder approach, which is another implementation possibility independent of the time-segment processing approach from the previous chapter. Their was a clear need for a clarification of the phase vocoder approach in this domain. Though it has been known for years, we have provided a general framework and simple implementations upon which more complex effects may be built. Some of them can reduce the phasiness of the process or perform special high-level processing on transients. Other digital audio effects have been described that fit well under the name “mutations.” They are based on modifying the magnitude and phase of one or two time-frequency representations. They put a special flavor on sounds, which musicians characterize as granulation, robotization, homogenization, purification, metallization and so on. Once again, the goal of this chapter is to give a general framework and unveil some of the basic implementations of these alterations of sound, which can be extended to more complex modifications at will.
As a final remark, one can say that no digital audio effect, and time-frequency processing in particular, would exist without a sound. Only a good adaptation of the sound with the effect can give rise to musical creativity. This is the reason why some of the basic algorithms presented put in the hands of creative musicians and artists can give better results than much more complex algorithms in the hands of conventional persons.