On calibration and coherence signal analysis of the CHIL microphone network at IRST

The purpose of this work is to describe the Microphone Network presently used at ITC-irst for multi-microphone data collection and prototype development, with the specific aim of conducting research inside the CHIL European Project. In the project, we define a generic multi-sensor system which consists of two main components: a distributed multi-camera system for visual room observation, including several calibrated cameras, and a multi-microphone system for acoustic scene analysis, which consists of microphone arrays, microphone clusters, table top microphones and close-talking microphones allowing detection of multiple acoustic events, voice activity detection, ASR and speaker location and tracking [1]. The target scenario comprises seminars and meetings. The entire audio acquisition system is based on a common sampling rate of 44.1 kHz and a sample accuracy of 24 bit. Also for acoustic sensors, a detailed characterization process as well as a calibration step are necessary, according to the purpose of having a jointly consistent description of the audio-video sensor geometry. In the CHIL room at ITC-irst (See Figure 1), seven T-shaped microphone arrays, each consisting of four omnidirectional microphones, were installed in order to obtain an optimal coverage of the environment for speaker localization and tracking purposes. Moreover, a NIST-MarkIII array [2] of 64 microphones was installed on the wall facing the seminar speaker. Through its use the primary objective is far field ASR: however, benefits are expected also for what concerns the use of MarkIII signals for speech activity detection and speaker localization.