Detection and localization of partial audio matches
Abstract
Within recent years, several applications have emerged which require detection and accurate localization of unknown partial audio matches within a dataset. This requirement cannot be adequately addressed with state-of-the-art matching approaches based on fingerprinting. We propose a new approach that supports partial matching detection and localization within a dataset, and we evaluate it against a popular audio matching algorithm, showing that it performs significantly better for the given problem domain.
