Beyond Acoustic Sparsity and Linguistic Bias: A Prompt-Free Paradigm for Mispronunciation Detection and Diagnosis

Haopeng, Longfei, Xi, Haitong, Daisuke, Nobuaki · 2026

Read Beyond Acoustic Sparsity and Linguistic Bias: A Prompt-Free Paradigm for Mispronunciation Detection and Diagnosis by Haopeng, Longfei, Xi, Haitong, Daisuke, Nobuaki (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-2604-22133.txt.

Description

Geng and colleagues offer a prompt-free framework for spotting and diagnosing mispronunciations, splitting sound faithfulness from set guidance. First, CROTTC, a sound model, forces step-by-step, frame-level alignment to catch speech slips in fine detail. Second, the IF trick quietly feeds in slip knowledge under a transfer rule. CROTTC-IF scores 71.77 percent F1 on L2-ARCTIC and about 72 percent on Iqra'Eval2, showing that splitting sound from set priors gives strong detection.

Catalog record

CREATORHaopeng, Longfei, Xi, Haitong, Daisuke, Nobuaki
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE49823
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.22133

Browse the full library · About the Tower of Babel