Errata

This page serves as a collection of errata in works that I have published. It is organized by paper/work, and the errtum/errata is/are listed after the paper.

• Kelley, M. C., Perry, S. J., & Tucker, B. V. (2024). The Mason-Alberta Phonetic Segmenter: A forced alignment system based on deep neural networks and interpolation. Phonetica 81(5), 451-508. https://doi.org/10.1515/phon-2024-0015.

  • Line 11 of Algorithm 1 says that "backtracking" should be used on M. What was meant was that "backtracing" should be used on M. The results were not affected since backtracing is what was actually performed for the alignment.

• Kelley, M. C., & Tucker, B. V. (2018). A comparison of input types to a deep neural network-based forced aligner. Proc. Interspeech 2018, 1205-1209.

  • In the Algorithm 1 on page 1206, lines 19-24 specify using an argmax function and backward-traversing the result to find the alignment. This is not guaranteed to find the optimal path that led to the last cell in the dynamic programming matrix. Instead, a standard backtracking algorithm should be used, considering [i, j-1] and [i-1, j-1] (but not [i-1, j]) as potential progenitors of the current cell.