Tag
This paper introduces a scalable method using acoustic masking and ASR models to measure consonant contributions to word intelligibility, validating it against phoneme frequency and functional load across four languages.