Downloads: 118 | Views: 213

Research Paper | Electronics & Communication Engineering | India | Volume 4 Issue 1, January 2015

Performance Analysis of RPCA Algorithm for Segregation of Singing Voice from Polyphonic Music

Priyanka K. Umap | Kirti B. Chaudhari

Abstract: There are numerous real world applications for singing voice segregation from mixed audio. Using Robust Principal Component Analysis which is a compositional model for segregation, which decomposes the varied source of audio signal into low rank and sparse components, where it is assumed that musical accompaniment as low rank subspace since musical signal model is repetitive in character while singing voices can be treated as moderately sparse in nature within the song. Performance evaluation of the algorithm is verified by various performance measurement parameter such as source to distortion ratio (SDR), source to artifact ratio (SAR), source to interference ratio (SIR) and Global Normalized source to Distortion Ratio GNSDR.

Keywords: Robust Principle Component Analysis RPCA, Augmented Lagrange Multiplier ALM, low rank matrix, sparse matrix, Singing voice separation

Edition: Volume 4 Issue 1, January 2015,

Pages: 2582 - 2586