Timbre-based machine learning of clustering Chinese and Western Hip Hop music

Abstract

Chinese and Western Hip Hop musical pieces are clustered using timbre-based Music Information Retrieval (MIR) and machine learning (ML) algorithms. Psychoacoustically motivated algorithms extracting timbre features such as spectral centroid, roughness, sharpness, sound pressure level (SPL), flux, etc. were extracted form 38 contemporary Chinese and 38 Western 'classical' (USA, Germany, France, Great Britain) Hip Hop pieces. All features were integrated over the pieces with respect to mean and standard deviation. A Kohonen self-organizing map, as integrated in the Computational Music and Sound Archive (COMSAR\cite{COMSAR}) and apollon\cite{apollon} framework was used to train different combinations of feature vectors in their mean and standard deviation integrations. No mean was able to cluster the corpora. Still SPL standard deviation perfectly separated Chinese and Western pieces. Spectral flux, sharpness, and spread standard deviation created two sub-cluster within the Western corpus, where only Western pieces had strong values there. Spectral centroid std did sub-cluster the Chinese Hip Hop pieces, where again only Chinese pieces had strong values. These findings point to different production, composition, or mastering strategies. E.g. the clear SPL-caused clusters point to the loudness-war of contemporary mastering, using massive compression to achieve high perceived loudness.

Modelling the perception of music in brain network dynamics

“…The characteristic of contrasting parts can be revealed not only by music analysis using pen and paper but also by different computational methods by the music information retrieval discipline, like the amplitude of a piece of music that corresponds to the subjective perception of loudness. Also other properties of the stimulus, such as the spectral centroid that corresponds to the perceived brightness of a sound, or the fractal correlation dimension (Grassberger and Procaccia, 1983a,b) corresponding to the perceived density and thereby representing the complexity of a piece of music, are drivers of the musical form (Bader, 2013;Hartmann and Bader, 2020;Bader, 2021;Bader et al, 2021;Linke et al, 2021).…”

Influence of music on empirical brain networks

“…The characteristic of contrasting parts can be revealed not only by music analysis using pen and paper but also by different computational methods by the music information retrieval discipline, such as the amplitude of a piece of music that corresponds to the subjective perception of loudness. Also other properties of the stimulus, such as the spectral centroid that corresponds to the perceived brightness of a sound, or the fractal correlation dimension that corresponds to the perceived density may also contribute to the emergence of musical form (Bader, 2013;Hartmann and Bader, 2020;Bader, 2021;Lindenbaum et al, 2021;Bader et al, 2021).…”

Acoustic features from the recording studio for Music Information Retrieval Tasks

“…Music Information Retrieval (MIR) task reach from blind source separation over automatic music transcription to music recommendation. Musicologists often use acoustic features and machine learning methods as explorative tool to analyze ethnographical recordings [4,6], popular music [2,5] and large music collections [2,3]. One of the most-used acoustic features for music analysis are Mel-Frequency Cepstral coefficients (MFCCs), even though they neither correlate with any aspect of music perception nor music production and mixing [8].…”