Part of Advances in Neural Information Processing Systems 9 (NIPS 1996)
Te-Won Lee, Anthony Bell, Russell Lambert
We address the difficult problem of separating multiple speakers with multiple microphones in a real room. We combine the work of Torkkola and Amari, Cichocki and Yang, to give Natural Gra(cid:173) dient information maximisation rules for recurrent (IIR) networks, blindly adjusting delays, separating and deconvolving mixed sig(cid:173) nals. While they work well on simulated data, these rules fail in real rooms which usually involve non-minimum phase transfer functions, not-invertible using stable IIR filters. An approach that sidesteps this problem is to perform infomax on a feedforward archi(cid:173) tecture in the frequency domain (Lambert 1996). We demonstrate real-room separation of two natural signals using this approach.