LIWHIZ: A NON-INTRUSIVE LYRIC INTELLIGIBILITY PREDICTION SYSTEM FOR THE CADENZA CHALLENGE

Ram C. M. C. Shekar; Iván López-Espejo

ICASSP 2026poster0 citations

LIWHIZ: A NON-INTRUSIVE LYRIC INTELLIGIBILITY PREDICTION SYSTEM FOR THE CADENZA CHALLENGE

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.

BibTeX

@inproceedings{icassp2026_liwhizanonintrus,
  title = {LIWHIZ: A NON-INTRUSIVE LYRIC INTELLIGIBILITY PREDICTION SYSTEM FOR THE CADENZA CHALLENGE},
  author = {Ram C. M. C. Shekar and Iván López-Espejo},
  booktitle = {ICASSP 2026},
  year = {2026}
}