首页>
外国专利>
Sound content acknowledge model i.e. extractor, creating method for detecting sound event from e.g. vehicle impact noise, involves decomposing problem into two problems when descriptor does not attains threshold
Sound content acknowledge model i.e. extractor, creating method for detecting sound event from e.g. vehicle impact noise, involves decomposing problem into two problems when descriptor does not attains threshold
The method involves automatically testing relevance of audio descriptors e.g. sound source descriptors. Target and/or an antagonist classes are defined by decomposing one class into sub-assemblies defined in the classification of the class. Data bases (B2 - B4) are developed from the classes along the iteration of the defining step until one descriptor attains discrimination threshold for a sound identification problem corresponding to the data bases. The defining and developing steps are used for decomposing the problem into two problems when each descriptor does not attain the threshold. An independent claim is also included for a system for automatically creating a sound content acknowledge model.
展开▼
