| 摘要 |
A method comprises determining acoustic feature(s) of audio data, generating first and second classifications based on the feature(s) using first and second classifiers respectively, generating at least one third classification based on said first and second classifications using a third classifier and storing tag(s)for said audio data based on said third classification. The first and/or third classifiers may be non-probabilistic, e.g. a support vector machine (SVM) classifier. The second classifier may be probabilistic, e.g. based on a Gaussian Mixture Model (GMM). Another method determines whether audio data matches an audio track in a catalogue, based on audio-fingerprints and/or metadata. If so, information for the audio data is obtained from the matching track. If not, then one or more acoustic features of the audio data are extracted and used to continue the search. If no match is found, then information based on the extracted features is uploaded to the catalogue. |
