Real-time Speech Enhancement System for Surgical Systems

* Presenting author
Day / Time: 20.03.2019, 09:00-09:20
Room: Saal 5-a
Typ: Regulärer Vortrag
Abstract: The communication in surgery rooms might be difficult, especially if the surgeon is located in some distance from the patient due to the usage of minimally invasive operation devices. In order to improve the communication between the medical staff around the patient table and the surgeon controlling the operation device dedicated signal processing can be applied. In particular, such speech enhancement systems focus on suppression of the stationary background noise caused by medical equipment, suppression of the feedback occurring due to a closed electro-acoustic loop, and lastly suppression of instationary noise, e.g. sharp clanging noises. All this should be done without compromising the quality and intelligibility of the speech signals.In this work, we present an overview of a generic speech enhancement system comprising mainly of a traditional Wiener filter for background noise suppression, a model-based feedback suppression along with additional stages for multidirectional communication and non-linear time domain processing for the attenuation of instationary noise. All this was implemented and tested on an embedded hardware platform with respect to the special requirements of a surgery room.