Technical Program

Paper Detail

Paper:MLSP-P5.1
Session:Machine Learning for Speech and Audio Applications
Location:Poster Area 9
Session Time:Friday, May 27, 09:30 - 11:30
Presentation Time:Friday, May 27, 09:30 - 11:30
Presentation: Poster
Topic: Machine Learning for Signal Processing: Speech and Audio Processing Applications
Paper Title: AUDIO SOURCE SEPARATION BY BASIS FUNCTION ADAPTATION
Authors: Yinyi Guo, Mofei Zhu, Stanford University, United States
Abstract: The problem of audio source separation from a monophonic sound mixture having known instrument types but unknown timbres is presented. An improvement to the Probabilistic La- tent Component Analysis (PLCA) source separation method is proposed. The technique uses a basis function dictionary to produce a first round PLCA source separation. The PLCA weights are then refined by incorporating note onset informa- tion. The source separation is then performed using a sec- ond round PLCA in which the refined weights are held fixed, and the basis functions are updated. Preliminary experimen- tal results on mixtures of two instruments are quite promising, showing a 6 dB improvement in SIR over standard PLCA.