Additive analysis-synthesis is a powerful way to completely parameterize a sound event into sinusoidal partials with their frequencies, amplitudes, and phases. This benefit is also its curse: It puts every minute detail of a sound event at our disposal, but leaves us with the task to control and manipulate this mass of parameters in a sensible way. So far, control is done by specifying the change of every single parameter over time by break point functions (see section 4.4). Since the number of partials can easily rise into the hundreds, modifications are tedious. Moreover, doing valid manipulations in regard of signal processing and from a musical perspective is not obvious, and, what's more, the parameters are interdependent (e.g. changing the frequency of the partials changes the spectral envelope, often with undesirable results, as shown in section 2.3.1).
In [FRD92] it is suggested to use spectral envelopes to control the amplitudes of the partials for resynthesis. This drastically reduces the number of parameters, provides us with parameter sets which are easily understandable (e.g. formants), and renders frequency and amplitude control independent from each other.
Also, for the residual noise, the modeling by filtering of a white noise with spectral envelopes (estimated with linear prediction) renders this component of sound accessible to manipulation. This has not been possible before in the sampled signal representation of the residual.
The most significant advantage, however, lies in the unified handling of noise and harmonic part, because the spectral envelope of the residual noise is represented in the same way as the spectral envelope of the sinusoidal part. Therefore, the very same manipulation can affect both parts synchronously, if this is desired [RDG95].