A two-input two-output speech enhancement method that preserves binaural spatial information in the output is preferred for realizing a comfortable auditory communication system. Such a system benefits from the noise reduction capability provided by signal processing technology, in addition to the binaural processing of the human auditory system. We investigated a two-input two-output speech enhancement method that calculates soft decision mask filters to attenuate a noisy time-frequency bin. The soft decision mask filter is estimated for each direction of arrival (DOA) based on a noise-target ratio calculated using an adaptive filter that cancels the target signal. Results of computer simulations show that the proposed method has superior capabilities for maintaining spatial information in the two-output signals and for segregating the target signal in arbitrary azimuth and elevation DOA.
|Number of pages
|Proceedings - European Conference on Noise Control
|Published - 2008
|7th European Conference on Noise Control 2008, EURONOISE 2008 - Paris, France
Duration: 2008 Jun 29 → 2008 Jul 4