fastabx: A library for efficient computation of ABX discriminability
This provides a practical tool for researchers in representation learning, especially beyond speech processing, though it is incremental as it addresses a gap in existing software rather than proposing new methods.
The authors tackled the lack of efficient tools for building ABX discrimination tasks, a measure used to evaluate category separation in representations, by introducing fastabx, a high-performance Python library that enables rapid development and distance calculations.
We introduce fastabx, a high-performance Python library for building ABX discrimination tasks. ABX is a measure of the separation between generic categories of interest. It has been used extensively to evaluate phonetic discriminability in self-supervised speech representations. However, its broader adoption has been limited by the absence of adequate tools. fastabx addresses this gap by providing a framework capable of constructing any type of ABX task while delivering the efficiency necessary for rapid development cycles, both in task creation and in calculating distances between representations. We believe that fastabx will serve as a valuable resource for the broader representation learning community, enabling researchers to systematically investigate what information can be directly extracted from learned representations across several domains beyond speech processing. The source code is available at https://github.com/bootphon/fastabx.