Estimating the short-time bandwidth of wheeze sounds

This paper proposes a system for estimating the short-time bandwidth of wheeze sounds using a network of coupled adaptive bandpass filters. Inherent spectral characteristics of the input signal (wheeze) are exploited to determine its bandwidth. The bandwidth information can be used to reconfigure the overall power consumption of components such as the preamplifier, anti-aliasing filter, drivers and ADC of a wheeze detection system. The simulation results in AMI 0.5 μm with Vdd of 3.3V are presented. The proposed system consumes 34 μW, for a input signal bandwidth of 10kHz. This implementation can be extended to other audio sensor interface systems resulting in bandwidth dependent power consumption.