The CUIDADO Project

The CUIDADO Project (Content-based Unified Interfaces and Descriptors for Audio/music Databases available Online) aims at developing a new chain of applications through the use of audio/music content descriptors, in the spirit of the MPEG-7 standard. The project includes the design of appropriate description structures, the development of extractors for deriving high-level information from audio signals, and the design and implementation of two applications the Sound Palette and the Music Browser. These applications include new features, which systematically exploit high-level descriptors and provide users with content-based access to large catalogues of audio/music material. The Sound Palette focuses on audio samples and targets professional users, whereas the Music Browser addresses a broader user target through the management of Popular music titles. After a presentation of the project objectives and methodology, we describe the original features of the two applications based on the systematic use of descriptors and the technical architecture framework on which they rely.

However, in order to come up to effective results, several constraints, which form the basic methodological framework of the project, must be taken into account

i) the necessity of a top-down approach, aiming at deriving high-level descriptors, which form the basic knowledge structures accessible by the user in target applications. These knowledge structures must be adapted to application functions and consistent with the user cognition, i.e. the way his own knowledge about sound and music is organized.

ii) the interest in combining this top-down approach with a bottom-up approach, which extracts low-level descriptors from an automatic analysis of the audio signal. The objective is to automatically compute required high-level descriptors, avoiding manual input when possible. The main related issues, discussed in [(Herrera, P. Peeters, G. Dubnov, S. 2003), (Herrera, P. Yeterian, A. Gouyon, F. 2002)] in the context of sound classification, are how to choose a set of relevant low-level descriptors and how to map targeted high-level descriptors onto them, by using appropriate machine learning techniques.