Ad-hoc mobile array based audio segmentation using latent variable stochastic model

Chetupalli, SR and Bhowmick, A and Sreenivas, TV (2019) Ad-hoc mobile array based audio segmentation using latent variable stochastic model. In: 27th European Signal Processing Conference, EUSIPCO 2019, 2 - 6 September 2019, Coruna; Spain.

PDF ESPC_2019.pdf - Published Version Restricted to Registered users only Download (12MB) | Request a copy

Abstract

Segmentation/diarization of audio recordings using a network of ad-hoc mobile arrays and the spatial information gathered is a part of acoustic scene analysis. In this ad-hoc mobile array network, we assume fine (sample level) synchronization of the signals only at each mobile node and a gross synchronization (frame level) across different nodes is sufficient. We compute spatial features at each node in a distributed manner without the overhead of signal data aggregation between mobile devices. The spatial features are then modeled jointly using a Dirichlet mixture model, and the posterior probabilities of the mixture components are used to derive the segmentation information. Experiments on real life recordings in a reverberant room using a network of randomly placed mobile phones has shown a diarization error rate of less than 14 even with overlapped talkers. © 2019,IEEE

Item Type:	Conference Paper
Publication:	European Signal Processing Conference
Publisher:	European Signal Processing Conference, EUSIPCO
Additional Information:	The copyright for this article belongs to European Signal Processing Conference, EUSIPCO
Keywords:	Audio acoustics; Mixtures; Mobile computing; Reverberation; Sensor networks; Signal processing; Stochastic models; Stochastic systems, Acoustic scene analysis; Acoustic sensor networks; Diarization; Dirichlet distributions; Dirichlet mixture model; Posterior probability; Response power; Segmentation informations, Mobile ad hoc networks
Department/Centre:	Division of Electrical Sciences > Electrical Communication Engineering
Date Deposited:	06 Jan 2023 09:41
Last Modified:	06 Jan 2023 09:41
URI:	https://eprints.iisc.ac.in/id/eprint/78836

Actions (login required)

View Item