Figure 5
From: A dynamic multi-channel speech enhancement system for distributed microphones in a car environment
Processed signal spectra for different processings based on an example scenario. Noisy microphone spectra for each of M=4 channels (first two rows) and the output signals for Q=2 are depicted. Three different processings are shown: The processed two output signal spectrograms without ISC and RCS (third row), the processed signal with ISC but again without RCS (fourth row), and finally the outputs based on the overall processing including ISC as well as RCS (last row). The speakers are active one after another (driver, front passenger, rear left passenger, rear right passenger). Noise is superposed from a car driving at 80 km/h. Each passenger speaks at an SNR of 0 dB.