Speech enhancement for hands-free terminals

This paper discusses signal processing methods for speech extraction in use with voice communication applications such as personal digital assistants (PDA), mobile telephone terminals and personal computers. The user is distant from the device and thus the speech signal entering the device may be subject to reverberation and may be disturbed by background noise. The proposed structure consists of a microphone array which allows for techniques of directional processing. Thus it is able to enhance a desired speech source and suppress other speech sources in the background. Three different optimal beamforming methods are considered in a real world car hands-free environment: an optimal near-field array gain optimization procedure, a theoretical diffuse noise field model for a point source and a least squares solution.