Musical Instrument Timbres Classification with Spectral Features
- PDF / 797,851 Bytes
- 10 Pages / 600 x 792 pts Page_size
- 1 Downloads / 188 Views
Musical Instrument Timbres Classification with Spectral Features Giulio Agostini Dipartimento di Scienze dell’Informazione, Universit`a degli Studi di Milano, Via Comelico 39, 20135 Milano, Italy Email: [email protected]
Maurizio Longari Dipartimento di Scienze dell’Informazione, Universit`a degli Studi di Milano, Via Comelico 39, 20135 Milano, Italy Email: [email protected]
Emanuele Pollastri Dipartimento di Scienze dell’Informazione, Universit`a degli Studi di Milano, Via Comelico 39, 20135 Milano, Italy Email: [email protected] Received 10 May 2002 and in revised form 29 August 2002 A set of features is evaluated for recognition of musical instruments out of monophonic musical signals. Aiming to achieve a compact representation, the adopted features regard only spectral characteristics of sound and are limited in number. On top of these descriptors, various classification methods are implemented and tested. Over a dataset of 1007 tones from 27 musical instruments, support vector machines and quadratic discriminant analysis show comparable results with success rates close to 70% of successful classifications. Canonical discriminant analysis never had momentous results, while nearest neighbours performed on average among the employed classifiers. Strings have been the most misclassified instrument family, while very satisfactory results have been obtained with brass and woodwinds. The most relevant features are demonstrated to be the inharmonicity, the spectral centroid, and the energy contained in the first partial. Keywords and phrases: timbre classification, content-based audio indexing/searching, pattern recognition, audio features extraction.
1.
INTRODUCTION
This paper addresses the problem of musical instrument classification from audio sources. The need for this application strongly arises in the context of multimedia content description. A great number of commercial applications will be available soon, especially in the field of multimedia databases, such as automatic indexing tools, intelligent browsers, and search engines with querying by content capabilities. The goal of automatic music-content understanding and description is not new and it is traditionally divided into two subtasks: pitch detection, or the extraction of score-like attributes from an audio signal (i.e., notes and durations), and sound-source recognition, or the description of sounds involved in an excerpt of music [1]. The former has received a lot of attention and some recent experiments are described in [2, 3]; the latter has not been studied so much because of the lack of knowledge about human perception and cognition of sounds. This work belongs to the second area and it is devoted to a more modest goal, but important
nevertheless, automatic timbre classification of audio sources containing no more than one instrument at a time (source must be monotimbral and monophonic). Focusing on this area, the forthcoming MPEG-7 standard should provide a list of metadata for multimedia content [4], nevertheless, two important as
Data Loading...
