Pattern Recognition in Music

Sammendrag/Abstract: This report gives a brief overview of different applications, problems and methods related to pattern recognition in music. Many of the applications of musical pattern recognition are connected to music information retrieval. This area covers fields like information retrieval, signal processing, pattern recognition, artificial intelligence, databases, computer music and music cognition. The report focuses on problems and methods related to signal processing and pattern recognition. Automatic music transcription and content-based music retrieval are the problems that have received the most attention within this area. For music transcription the current state-of-the-art is that monophonic transcription for well-defined musical instruments has to a large degree been solved as a research problem, while transcription of polyphonic music remains a research issue for the general case. Content-based retrieval based on audio queries is somewhat dependent on the transcription, although a full transcription may not be necessary to find similarity. Other problems like genre classification, music summarization and musical instrument recognition are also treated in the report. These are also problems that are related to music retrieval in that these techniques can be used for organizing the music databases and to present the results to users. Less research has been done in these areas.