PathBench: Speech Intelligibility Benchmark for Automatic Pathological Speech Assessment
Bence Mark, Thomas, Defne, Tomoki · 2026
Read PathBench: Speech Intelligibility Benchmark for Automatic Pathological Speech Assessment by Bence Mark, Thomas, Defne, Tomoki (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-08097.txt.
Description
Halpern and colleagues offer PathBench, a shared yardstick for scoring pathological speech. Until now, comparison was hard because work was split across private data sets with uneven rules. PathBench instead draws on public data and tests methods that need no reference, a reference text, or reference audio, under three setups: Matched Content, Extended, and Full. The setups mirror a linguist using controlled prompts versus a machine learning specialist using as much data as possible. Baselines span six data sets, with DArtP topping reference-free methods on average correlation.
Catalog record
| CREATOR | Bence Mark, Thomas, Defne, Tomoki |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 25672 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.08097