Today sees the unveiling of the world’s first online music recognition system. Omras (online music recognition and searching), has been developed by a team of US and UK scientists and who claim it enables computers to recognise complex pieces of music when they are played or hummed.
The British scientists are based at Queen Mary, University of London. The team says that the system will allow internet users to search for polyphonic audio, which means they can retrieve recordings from a full orchestra or complex piano chord, rather than a sound made by a single instrument or voice. It can identify music from a snippet of a tune or even from someone humming the melody.
“The system can recognise music played by a seven-year-old child on a piano. For example, the child could play My Way and even though it sounds completely different to the Frank Sinatra version, the system can still recognise it and list all the matches,” explains Kate Hunter, from Queen Mary, University of London.
Because the recognition system is based on mathematic algorithms, Hunter claims it is completely accurate. But you can also use the system to find variations on a tune — for example, there are 12 versions of Twinkle, Twinkle Little Star, and if you play one it will identify them all.
But Omras’ talents don’t end there, it can also generate a score for the music once it has found it online. Hunter says it can help to counter plagiarism and sampling of music that is still protected by copyright as it can identify underlying similarities between pieces. “It can make comparisons between notes and scores” to illustrate how heavily one tune is based on another.
Currently the system is at a very early stage: “It is a basic search engine based on 3,000 pieces of classical music that are not copyrighted – we can’t afford pop music,” explains Hunter. The team are hoping to interest a commercial music publisher in the scheme so they can extend its possibilities, allowing users to search for their favourite chart hits.
Hunter says one way in which a commercial publisher could benefit is by charging users to download scores generated from their music by the system.
The system is currently running on a specially adapted PC, but Hunter says there is no reason it couldn’t run on standard computers in the future. Whether or not we will have a musical search engine along the lines of Google depends on whether the team can attract investors, says Hunter. However, she said it can now be used in an academic environment, to allow music students to search for non-copyrighted classical melodies.