Design of a subspace tracking based speech enhancement system

In this paper, we describe a design of the signal subspace speech enhancement based on subspace tracking algorithm. The proposed architecture incorporates a perceptual filterbank, which is derived from psycho-acoustic model for subband processing. Considering hardware solutions to the subspace tracking algorithm, we derive a VLSI architecture of the algorithm without delayed updating. The experiments were performed using the TAICAR in-car noisy speech database. Simulation results are presented to validate our algorithm and hardware architectures.