Arabic music analysis using artificial intelligent teqniques

الملخص EN

Singing is the use of the human voice to make musically meaningful sounds, and it is used in most cultures for enjoyment or self-expression.

songs are audio signal and musical instrument representations.

speech, background noise, and music able to be identified by an audio signal analysis and separation system.

the singing voice in a song provides useful information on pitch range, music content, music pace, and rhythm.

nowadays, with multimedia technology, there are many audio editing software available as well as audio merging software by mixing singing voice and music together, but most of these applications are in the field of western music and there is a less application in the field of Arabic music as well as its not free.

one of the primary qualities that identify music based on a specific set of patterns is genre.

however, the genres of Arabic music on the web are loosely defined, making automatic classification of Arabic audio genres difficult.

in this paper, a system has been proposed that consider a form of Arabic music analysis and classification which is a two-stage based system: the first stage is separation between Arabic singing voice and melody using CRPCA which is an extension of RPCA in our previous work.

then extracting the Arabic musical genres which forms musical melody that had extracted from the previous process.

MEL frequency cepstral coefficients and pitch used to extract feature from musical signal, then supports vector machine algorithm used for classification process.

the experimental results show that CRPCA can achieve greater separation performance than earlier approaches, particularly when using temporal frequency masking.

furthermore, the duration of operation on CRPCA under the same conditions, is shorter than others, as well as the most important benefit of the separation improvement that its in fact of improving the classification and analysis process in the next stages.