This paper proposes a blind source separation (BSS) method to improve the performance of the degenerate unmixing estimation technique (DUET) when sound sources are recorded using closely spaced stereo microphones. In particular, the attenuation-delay-based discrimination analysis employed in DUET is replaced with a microphone spacing- and source direction-based discrimination analysis in order to remedy the problem of DUET when the attenuation factors between recorded stereo audio signals are not distinguishable. In other words, the proposed BSS method generates a histogram as a function of the microphone spacing and the directional difference between stereo signals. Next, the generated histogram is used to partition the time-frequency representations of the mixtures into that of each sound source. The performance of the proposed method is evaluated by means of both objective and subjective measures. Consequently, it is shown from the evaluation that the proposed BSS method outperforms the conventional DUET in a closely spaced stereo microphone recording environment.
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.