Jul 2026
Probing Low-Level Acoustic Attribute Encoding in CLAP Audio Embeddings
This work analyzes CLAP audio embeddings through a probing framework, studying the encoding of three fundamental perceptual dimensions: reverberation, loudness, and spectral content, measured via spectral centroid (SC) and relative pitch (RP).
Héctor Martel, Joe Hennessy-Priest, Taemin Cho
· arXiv.org · 1 citation