| 摘要 |
<p>An audio fingerprint feature-based music retrieval system, comprising a preprocessing module, a feature extraction module, a reverse index module and a precise matching module. The preprocessing module primarily conducts conversion, resampling, and filtering of audio signals; the feature extraction module is a representation of an audio file employing audio fingerprint features and selecting the most stable points in the frequency spectrum as feature points via double screening based on a dynamic threshold, and using a pair of points to represent a feature; the reverse index module uses the features as keywords, and establishes a reverse index by the features in a music library, and returns an index result according to the number of the same keywords; the precise matching module combines the timing relationships between audio features and employs improved editing distance as the degree of similarity between two feature sequences, so as to optimize the index result.</p> |
