Multi-source TDOA estimation in reverberant audio using angular spectra and clustering

Abstract

-source TDOA estimation in reverberant audio using angular spectra and clustering. Signal Processing, Elsevier, 2012Elsevier, , 92, pp.1950Elsevier, -1960Elsevier, . <10.1016Elsevier, /j.sigpro.2011 Multi-source TDOA estimation in reverberant audio using angular spectra and clustering AbstractWe consider the problem of estimating the time differences of arrival (TDOAs) of multiple sources from a two-channel reverberant audio signal. While several clustering-based or angular spectrum-based methods have been proposed in the literature, only relatively small-scale experimental evaluations restricted to either category of methods have been carried out so far. We design and conduct the first large-scale experimental evaluation of these methods and investigate a two-step procedure combining angular spectra and clustering. In addition, we introduce and evaluate five new TDOA estimation methods inspired from signal-to-noise-ratio (SNR) weighting and probabilistic multi-source modeling techniques that have been successful for anechoic TDOA estimation and audio source separation. The results show that clustering-based methods do not improve upon angular spectrum-based methods. For 5 cm microphone spacing, the best TDOA estimation performance is achieved by one of the proposed SNR-based angular spectrum methods. For larger spacing, a variant of the generalized cross-correlation with phase transform (GCC-PHAT) method performs best.

A 3D model for room boundary estimation

“…To localize the source, two different parameters can be used: the sound time of arrivals (TOAs) (or similarly time difference of arrivals (TDOAs) between two different microphones) which can be estimated by the generalized crosscorrelation with phase transform (GCC-PHAT) method [1] This work was supported in the UK by the Engineering and Physical Sciences Research Council (EPSRC) Grants EP/K014307/1 and EP/L000539/1, and the MOD University Defence Research Collaboration in Signal Processing. [2], and the direction of arrivals (DOAs) through the spatial method proposed in [3].…”

Spatial Sound Localization via Multipath Euclidean Distance Matrix Recovery

“…Similar to (12), a third-order polynomial is obtained with maximum three roots and the one which globally minimizes the cost function is chosen. Hence, the optimization 3 Torgerson's double centering [28] as implemented in step 4 of Algorithm 1, is subtracting the row and column means of the matrix from its elements, adding the grand mean and multiplying by -1/2. The double centered matrix is scalar products relative to the origin and the coordinates is determined by the singular value decomposition (steps 5-6).…”

“…The generalized cross-correlation is typically used where the peak location of the cross-correlation function of the signal of two microphones is mapped to an angular spectrum for direction of arrival estimation [1]. A weighting scheme is often employed to increase the robustness of this approach to noise and multi-path effect [2,3]. The TDOA information can also be used to design a beamformer for directional sound acquisition.…”

Keyword Based Speaker Localization: Localizing a Target Speaker in a Multi-speaker Environment

“…Nevertheless, due to the approximate disjointness of speech signals in the time-frequency plane [23], most time-frequency bins are dominated by a single speaker. This property has motivated clustering-based localization algorithms, which iteratively identify the time-frequency bins dominated by each speaker and reestimate the corresponding DOAs [24][25][26]. It has also recently been exploited to design training data for multi-speaker DNN-based localization [18,19].…”