Pierre Hanna

Analysis method to approximate the spectral density of noises

Usual noise models represent sounds with filtered white noise, and extract only the spectral envelope from the sounds analyzed. Psychoacoustic experiments have shown the ability of humans to discriminate noise bands with different spectral densities. This property may be taken into account for the mathematical representation of noises. We propose an original method to approximate this new parameter. The limitations of the discrete short-term transform lead us to consider a new approach. This technique is based on statistical studies of the intensity fluctuations which are theoretically related to the spectral density. An algorithm is detailed and its parameters are discussed. Experiments show the quality and the limitations of this method. Although it may be improved, this estimation technique is accurate enough to be applied in a complete analysis/synthesis model for noisy sounds.

SATIN: A Persistent Musical Database for Music Information Retrieval

This paper introduces SATIN, the Set of Audio Tags and Identifiers Normalized. SATIN is a database of 400k audio-related metadata and identifiers that aims at facilitating reproducibility and comparisons among the MIR algorithms. The idea is to take advantage of partnerships between scientists and private companies that host millions of tracks. Scientists can send their feature extraction algorithm to companies along SATINs identifiers and retrieve the corresponding features. This procedure allows the MIR community to have access to more tracks for classification purposes. Afterward, scientists can provide to the MIR community the classification result for each track, which can then be compared with other algorithms results. SATIN thus resolves the major problems of accessing more tracks, managing copyrights locks, saving computation time, and guaranteeing consistency over research databases. We also introduce SOFT1, the first Set Of FeaTures extracted by a company thanks to SATIN.

Exploring the Complementarity of Audio-Visual Structural Regularities for the Classification of Videos into TV-Program Collections

This article proposes to analyze the structural regularities from the audio and video streams of TV-programs and explore their potential for the classification of videos into program collections. Our approach is based on the spectral analysis of distance matrices representing the short-and long-term dependancies within the audio and visual modalities of a video. We propose to compare two videos by their respective spectral features. We appreciate the benefits brought by the two modalities on the performances in the context of a K-nearest neighbor classification, and we test our approach in the context of an unsupervised clustering algorithm. These evaluations are performed on two datasets of French and Italian TV programs.

Segmentation of music video streams in music pieces through audio-visual analysis

Today, technologies for information storage and transmission allow the creation and development of huge databases of multimedia content. Tools are needed to facilitate their access and browsing. In this context, this article focuses on the segmentation of a particular category of multimedia content, audio-visual musical streams, into music pieces. This category includes concert audio-video recordings, and sequences of music videos such as the ones found in musical TV channels. Current approaches consist in supervised clustering in a few audio classes (music, speech, noise), and, to our knowledge, no consistent evaluation has been performed yet in the case of audio-visual musical streams. In this paper, we aim at estimating the temporal boundaries of music pieces relying on the assumed homogeneity of their musical and visual properties. We consider an unsupervised approach based on the generalized likelihood ratio to evaluate the presence of statistical breakdowns of MFCCs, Chroma vectors, dominant Hue and Lightness over time. An evaluation of this approach on 15 manually annotated concert streams shows the advantage of combining tonal content features to timbral ones, and a modest impact from the joint use of visual features in boundary estimation.

Recherche de documents musicaux par similarité mélodique

Dans un monde numerique en croissance permanente, la navigation dans de grandes bases de donnees musicales ne peut plus se limiter a une recherche par mot-cle textuel. Les nouvelles methodes de navigation reposent en grande partie sur l’estimation de la similarite musicale entre deux morceaux. Nous proposons d’adapter des methodes couramment employees dans le domaine de la Bioinformatique, notamment dans la recherche sur les sequences ARN, au contexte de la recherche d’informations musicales. Des evaluations a partir de bases de donnees reelles montrent que la precision de ces methodes est tout a fait satisfaisante. Nous presentons ensuite une application originale a la recherche automatique de plagiats. Les resultats d’experiences effectuees a partir de cas celebres de plagiats montrent alors un des interets de la comparaison de melodies.