The 1st Cadenza Lyric Intelligibility Prediction Challenge
Image by Sabena Costa from Pixabay
CLIP1 will be submitted as a proposal for an ICASSP 2026 SP Grand Challenge; however, the challenge will run regardless of its acceptance as an official ICASSP SP Grand Challenge.
To develop better music processing through machine learning, we need a reliable way to automatically evaluate audio quality. For music with lyrics, this includes a metric to assess the intelligibility of the sung words. Some songs are intrinsically less intelligible than others. Factors that can affect intelligibility include:
- Vocal style and articulation
- Song genre
- Mixing and production techniques
- Listener hearing ability
For speech technologies, intelligibility metrics have played a vital role in improving signal processing and machine learning, we hope the same can be achieved for music.
Challenge Overview
Participants will build models to predict lyrics intelligibility from audio recordings. The intelligibility metric would be derived from a predictive model that takes audio as input and estimates the score a listener would likely achieve in a listening test.
What will be provided?
- A dataset of song excerpts, some excerpts willl be provided as-is and others will be passed through a hearing loss simulator to mimic listeners with hearing loss but not wearing hearing aids.
- All samples will include lyrics intelligibility scores from listening tests.
- Software and baseline system.
Expressing interest
If you are interested in this challenge and intend to participate, please express your interest by filling out the form on the registration page. By doing so, you will help us better plan for a successful challenge.
You can also sign up to our Google group for alerts and discussions about our challenges.