Apparatus and method for automatic classification of audio signals

摘要

The present invention relates to an apparatus and a method for automatic classification of audio signals. Such an apparatus comprises: signal input means (3) for supplying audio signals; audio signal fragmenting means (4) for partitioning audio signals supplied by the signal input means (3) into audio fragments of a predetermined length; feature extracting means (5) for analysing acoustic characteristics of the audio signals comprised in the audio fragments; and classifying means (6) for discriminating the audio fragments provided by the audio signal fragmenting means (4) into a predetermined audio class based on predetermined audio class classsifying models (71,72,73) by using acoustic characteristics of the audio signals comprised in the audio fragments, wherein a predetermined audio class classifying model (71,72,73) is provided for each audio class and each audio class represents a respective kind of audio signals comprised in the corresponding audio fragment. It is a disadvantage that singing voice included in the audio signal frequently is misclassified as speech, particularly when the singing voice is the dominant signal component. The reason is that singing voice is more similar to speech than to music. To solve this problem, according to the present invention an individual predetermined audio class classifying model (71,72,73) is provided for at least each audio class "speech", "music" and "singing voice". Furthermore, the above disadvantage is overcome by the inventive method and the inventive software product.