Audio Blind Separation in real commercial music recordings is still an open problem. In the last few years some techniques have provided interesting results. This article presents a human-assisted clusterization of the DFT coefficients for the Time-Frequency Masking demixing technique. The DFT coefficients are grouped by adjacent panning and phase difference and an interactive graphical interface allows the user to select in real-time panning and phase difference ranges for each song. Results prove an implementation of such technique can be used to demix tracks from nowadays commercial songs. Sample sounds can be found here: http://www.iua.upf.es/~mvinyes/abs/demos
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.