Music analysis engine in the pipelines

German audio engineers are working on a music analysis engine which should automatically recognise and classify music genres worldwide.

A computer-based music analysis engine, which should automatically recognise and classify music genres worldwide, is under development in Germany.

If successful, the Global Music Database being compiled by the Franz Liszt Music Conservatory at Weimar, Germany will be able to show how pop stars like Shakira are often influenced by traditional music in crafting their hits.

According to Tiago de Oliveira Pinto from the school, an analysis of Shakira's music shows links to Cumbia, a style which originated in the musical and cultural fusion among native Colombians, slaves brought from Africa and the colonial-era Spanish.

The pilot project makes it possible in a short period to identify a single musical piece's historical and stylistic influences, for example the Brazilian samba of Roda, which first appeared in the state of Bahai but has its origins in Angola.

For the past two years, Brazilian music ethnologists have been working together with other experts from Europe, Asia, Latin America and Africa on a semantic search engine that automatically recognises basic musical attributes such as tone and rhythm.

"We are creating something that is independent of the global industry. The current search engines are only capable of finding identical musical pieces from huge databases," explained project worker Philip Kueppers.

"We synthesise basic elements from rhythm in order to deliver general musical information to users."

The concept focuses on the music itself rather than on a particular artist or song. This is a completely different form of measurement, meaning incorrect or incomplete results, such as confusing samba and salsa, are excluded.

The software was developed by musicologists working closely with digital audio engineers from Germany's Fraunhofer Institute.

"We presumed that when the musicologists gave us computer readable music, we would be able to recognise typical rhythmical patterns," explained Christian Dittmar, project leader for semantic music and technologies at the Fraunhofer.

Unfortunately, this failed to give the desired results.

"We therefore had to develop a system which was capable of comparing musical pieces directly," said Dittmar.

The advent of the internet, smartphones and tablets has completely changed the way music is listened to and sourced.

Source: AAP