K.R. Kale and J.G. Harris (USA)
Speech processing, Broadband beamforming, Narrow baseline, Frequency invariant beamformer
A beamforming method is developed to enhance voice quality and improve speech recognition accuracy for handheld devices. Stringent constraints for this algorithm include a low baseline, low complexity and low hardware requirements. The algorithm utilizes two microphones spaced 4 cm apart, small enough for operation on handheld devices. Method of implementing this frequency -independent algorithm is discussed and is compared to a far field frequency-dependent beamformer. Projected SNR improvements are simulated for both algorithms. Using only two microphones, this novel algorithm represents an excellent compromise among low hardware/software complexity, narrow baseline and SNR improvement.
Important Links:
Go Back
