Block diagram of the signal flow in Character.
Click to enlarge
Character employs mathematical modeling to extract parameters from the audio, which it applies to realtime adaptive filters. The output of the filters carry information about the source, and are fed back to the modeling unit in order to correct the adaptation over time. At the output stage, the characteristic components can be magnified according to the output of the filters, and the setting of the Character parameter.
Fast signal analysis plays a key role to this algorithm. The modeling is based on the assumption that audio is composed of a finite sum of sinusoidal waves with individual time-varying frequencies, phases, and amplitudes. The technique focuses on accurately describing the prominent spectral components, and their time-varying properties. Because Character is designed to accomodate a wide variety of sources, it constantly evaluates and corrects its own performance. This is necessary to protect the important information present in signal transients, and so the algorithm is capable of altering its settings as quickly as on a sample by sample basis.
The gains of the adaptive filters may change rapidly over time in correspondence with the signal properties. In addition to modifying the frequency response, Character can also adjust the dynamic aspects of a signal.
Therefore, Character is neither a static EQ, a Dynamic EQ, nor a Compressor or a Limiter. However, like these types of processors, Character also works with components already present in the signal that the user wants to bring out into the best characteristics of the sound.
In short, Character is a new color on your palette.
For a less scientific explanation of what Character does, please see the Character productpage.