This article investigates several parallelizable alternatives to DTW for estimating the alignment between two long sequences. Whereas most previous work has focused on reducing the total computation and/or memory costs of DTW, our focus is instead on reducing wall clock time by utilizing common hardware like GPUs that are optimized for parallel processing. We propose and study four different parallelizable alignment algorithms: the first three algorithms compute approximations of DTW by breaking the pairwise cost matrix into rectangular regions and processing the regions in parallel, and the fourth algorithm computes an exact DTW alignment by processing the cost matrix along diagonals rather than rows or columns. We characterize the performance of our proposed alignment algorithms on an audio-audio alignment task, and we develop GPU-based implementations for the two best-performing algorithms, which we call weakly-ordered Segmental DTW (WSDTW) and Parallelized Diagonal DTW (ParDTW). Our experiments indicate that ParDTW is the most practical and useful of the four algorithms: it computes an exact DTW alignment and reduces runtime by 1.5 to 2 orders of magnitude on long sequences compared to current alternatives. We present a comprehensive evaluation and study of the alignment accuracy, runtime, and practical limitations of the proposed alignment algorithms.
This paper investigates an ordered partial matching alignment problem, in which the goal is to align two sequences in the presence of potentially non-matching regions. We propose a novel parameter-free dynamic programming alignment method called hidden state time warping that allows an alignment path to switch between two different planes: a “visible” plane corresponding to matching sections and a “hidden” plane corresponding to non-matching sections. By defining two distinct planes, we can allow different types of time warping in each plane (e.g., imposing a maximum warping factor in matching regions while allowing completely unconstrained movements in non-matching regions). The resulting algorithm can determine the optimal continuous alignment path via dynamic programming, and the visible plane induces a (possibly) discontinuous alignment path containing matching regions. We show that this approach outperforms existing parameter-free methods on two different partial matching alignment problems involving speech and music.