Markus Koppenberger

Natural language processing of lyrics

We report experiments on the use of standard natural language processing (NLP) tools for the analysis of music lyrics. A significant amount of music audio has lyrics. Lyrics encode an important part of the semantics of a song, therefore their analysis complements that of acoustic and cultural metadata and is fundamental for the development of complete music information retrieval systems. Moreover, a textual analysis of a song can generate ground truth data that can be used to validate results from purely acoustic methods. Preliminary results on language identification, structure extraction, categorization and similarity searches suggests that a lot of profit can be gained from the analysis of lyrics.

Content-based music audio recommendation

We present the MusicSurfer, a metadata free system for the interaction with massive collections of music. MusicSurfer automatically extracts descriptions related to instrumentation, rhythm and harmony from music audio signals. Together with efficient similarity metrics, the descriptions allow navigation of multimillion track music collections in a flexible and efficient way without the need for metadata nor human ratings.

An industrial-strength content-based music recommendation system

We present a metadata free system for the interaction with massive collections of music, the MusicSurfer. MusicSurfer automatically extracts descriptions related to instrumentation, rhythm and harmony from music audio signals. Together with efficient similarity metrics, the descriptions allow navigation of multimillion track music collections in a flexible and efficient way without the need of metadata or human ratings.

The complex network of musical tastes

We present an empirical study of the evolution of a social network constructed under the influence of musical tastes. The network is obtained thanks to the selfless effort of a broad community of users who share playlists of their favourite songs with other users. When two songs co-occur in a playlist a link is created between them, leading to a complex network where songs are the fundamental nodes. In this representation, songs in the same playlist could belong to different musical genres, but they are prone to be linked by a certain musical taste (e.g. if songs A and B co-occur in several playlists, an user who likes A will probably like also B). Indeed, playlist collections such as the one under study are the basic material that feeds some commercial music recommendation engines. Since playlists have an input date, we are able to evaluate the topology of this particular complex network from scratch, observing how its characteristic parameters evolve in time. We compare our results with those obtained from an artificial network defined by means of a null model. This comparison yields some insight on the evolution and structure of such a network, which could be used as ground data for the development of proper models. Finally, we gather information that can be useful for the development of music recommendation engines and give some hints about how top-hits appear.

MTG-DB: a repository for music audio processing

Content-based audio processing researchers need audio and its related metadata to develop and test algorithms. We present a common repository of audio, metadata, ontologies and algorithms. We detail the hardware implementation, in the form of massive storage and computation cluster, the software and databases design and the ontology management of the current system. The repository, as far as copyright licenses allow, is open to researchers outside the music technology group to test and evaluate their algorithms.

Nearest-neighbor automatic sound annotation with a WordNet taxonomy

Sound engineers need to access vast collections of sound effects for their film and video productions. Sound effects providers rely on text-retrieval techniques to give access to their collections. Currently, audio content is annotated manually, which is an arduous task. Automatic annotation methods, normally fine-tuned to reduced domains such as musical instruments or limited sound effects taxonomies, are not mature enough for labeling with great detail any possible sound. A general sound recognition tool would require first, a taxonomy that represents the world and, second, thousands of classifiers, each specialized in distinguishing little details. We report experimental results on a general sound annotator. To tackle the taxonomy definition problem we use WordNet, a semantic network that organizes real world knowledge. In order to overcome the need of a huge number of classifiers to distinguish many different sound classes, we use a nearest-neighbor classifier with a database of isolated sounds unambiguously linked to WordNet concepts. A 30% concept prediction is achieved on a database of over 50,000 sounds and over 1600 concepts.