English

LIWhiz: A Non-Intrusive Lyric Intelligibility Prediction System for the Cadenza Challenge

Audio and Speech Processing 2026-02-02 v2 Sound

Abstract

We present LIWhiz, a non-intrusive lyric intelligibility prediction system submitted to the ICASSP 2026 Cadenza Challenge. LIWhiz leverages Whisper for robust feature extraction and a trainable back-end for score prediction. Tested on the Cadenza Lyric Intelligibility Prediction (CLIP) evaluation set, LIWhiz achieves a root mean square error (RMSE) of 27.07%, a 22.4% relative RMSE reduction over the STOI-based baseline, yielding a substantial improvement in normalized cross-correlation.

Keywords

Cite

@article{arxiv.2512.17937,
  title  = {LIWhiz: A Non-Intrusive Lyric Intelligibility Prediction System for the Cadenza Challenge},
  author = {Ram C. M. C. Shekar and Iván López-Espejo},
  journal= {arXiv preprint arXiv:2512.17937},
  year   = {2026}
}

Comments

Accepted to ICASSP 2026