An inventory was made of current, state-of-the-art/novel signal enhancement approaches. In an attempt to assess their suitability for application in future hearing devices the algorithms were evaluated and compared based on real-life data using perceptually weighted physical measures. Based on this evaluation a limited set of promising approaches was selected for real-time implementation on a common development platform. The selected algorithms were adapted to different auditory-profile subgroups and evaluated accordingly. Possible interaction effects between different classes of algorithms have been investigated.
The following types of signal enhancement procedures were considered in this study:
Single-channel noise reduction algorithms that can be used to suppress background noise (street noise, car noise, cafeteria noise, competing speaker signals, …).
Adaptive beamforming based noise suppression techniques that are intended for multi-microphone hearing instruments to combat background noise.
Blind source separation approaches that can be used in multi-microphone hearing instruments to separate different signal sources and to enhance the desired speaker signal.
Dereverberation algorithms that are designed to increase speech intelligibility and listening comfort in reverberant environments and diffuse background noise.
Feedback suppression techniques, which avoid the creation of acoustic feedback (whistling tones) due to an acoustic coupling between the hearing aid loudspeaker and microphone.