On the effect of snr and superdirective beamforming in speaker diarisation in meetings

Abstract

This paper examines the effect of sensor performance on speaker diarisation in meetings and investigates the use of more advanced beamforming techniques, beyond the typically employed delay-sum beamformer, for mitigating the effects of poorer sensor performance. We present superdirective beamforming and investigate how different time difference of arrival (TDOA) smoothing and beamforming techniques influence the performance of state-of-the-art diarisation systems. We produced and transcribed a new corpus of meetings recorded in the instrumented meeting room using a high SNR analogue and a newly developed low SNR digital MEMS microphone array (DMMA.2). This research demonstrates that TDOA smoothing has a significant effect on the diarisation error rate and that simple noise reduction and beamforming schemes suffice to overcome audio signal degradation due to the lower SNR of modern MEMS microphones.Index Terms-Speaker diarisation in meetings, digital MEMS microphone array, time difference of arrival (TDOA), superdirective beamforming

Development and evaluation of a digital MEMS microphone array for spatial audio

“…Recent works on digital microphone arrays discuss their implementation and compare their performance to analog arrays for various signal processing tasks, such as distant speech recognition [9], recognition of overlapping speech [10], speech enhancement [11], and speaker diarisation [12]. The work in [13] presents a digital array design for aeroacoustic measurements, while the implementation of a system for sound acquisition with MEMS microphones is presented in [14].…”

Detection of overlapped speech using lapel microphones in meeting

“…In recent years, meeting speech recognition (Maganti et al, 2007;Nasu et al, 2011) and meeting speaker diarization (Boakye et al, 2008;Ben-Harush et al, 2009;Stolcke et al, 2010;Sun et al, 2010;Valente et al, 2010;Vijayasenan et al, 2010;Boakye et al, 2011;Stolcke, 2011;Valente et al, 2011;Yella et al, 2011;Vijayasenan et al, 2012;Zwyssig et al, 2012) have been effectively utilized to transcribe and browse meeting procedures. However, their performance is usually low at the overlapped speech segments where more than one speaker is speaking.…”