Abstract: An object-oriented audio signal processing system is disclosed. The signal processing system is capable of converting virtually any sound resource into a format which can be used by associated sound hardware to produce a sound. A signal processing system incorporating the invention comprises one or more chains formed by a plurality of sound objects serially connected. A sound object is a specialized task which performs a particular operation such as, for example, sample rate conversion of a sound resource, decompression of a sound resource, mixing multiple sound resources or retrieving sound resources from a software application. The sound objects are preferably modular and use substantially the same calling conventions and interface. A chain of sound objects processes digital audio signals into an output which can be maintained in a storage area, e.g. memory or disk or played on sound hardware.

Abstract: Briefly, a method and apparatus for mixing a plurality of channels of digital audio samples using a non-linear clipping function which has a variable scaling factor is disclosed. The non-linear curving function is capable of being graphically represented by a curve which is sloped at the edges before reaching the maximum values. In a first embodiment, the non-linear clipping function is applied to the samples and values are calculated in real-time. Preferably, inputs to the non-linear clipping function are modified depending upon any significant trends, e.g. increasing or decreasing, in the sums of samples over time. In a second embodiment, to reduce processor overhead, prior to real time application of the non-linear clipping function, the calculations are performed and a lookup table is generated. This lookup table is then used during real time to apply the clipping function to the input data so that the non-linear function is not recalculated whenever multiple channels are mixed.