| 摘要 |
Signals (for example audio signals) which seem to be completely random, yet for which universally valid statements should be made, for example in the form of parameterizations which, on average, are accurate and can be determined only on the basis of short signal sections. Instead of simulating for example a Gaussian process, projections of algebraic operations-on the plane of real or complex numbers-of said signal sections for example are observed and proven for said astonishingly simple algebraic invariants. Said invariants are subsequently used as tags in order to perform, for example, a selection according to their frequency. On average, the present system proves to be more efficient than known methods to date. The practical-commercial application of said system covers nearly the entire signal processing field. The present document addresses in particular the stochastic observation of audio signals, as known for example from the field of digital audio broadcasting.
|
