Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
PHALAR: Phasors for Learned Musical Audio Representations
Davide Marincione, Michele Mancusi, Giorgio Strano, Luca Cerovaz, Donato Crisostomi, Roberto Ribuoli, Emanuele Rodolà
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add Learned Spectral Pooling instead of global average pooling (ablation) | PHALAR · MoisesDB K=64 | Top-1 accuracy: 51.97 -> 70.87 | Improved |
| add phase-equivariant complex head instead of magnitude only (ablation) | PHALAR · MoisesDB K=64 | Top-1 accuracy: 60.59 -> 70.87 | Improved |
| use phase-aware bilinear score instead of complex cosine similarity (ablation) | PHALAR · MoisesDB K=64 | Top-1 accuracy: 61.93 -> 70.87 | Improved |
| allow indefinite W instead of positive semi-definite W (ablation) | PHALAR · MoisesDB K=64 | Top-1 accuracy: 67.85 -> 70.87 | Improved |
| use general complex W instead of Hermitian W (ablation) | PHALAR · MoisesDB K=64 | Top-1 accuracy: 69.92 -> 70.87 | No clear effect |
| use CQT input instead of Mel-spectrogram (ablation) | PHALAR · MoisesDB K=64 | Top-1 accuracy: 69.21 -> 70.87 | Improved |
| use PHALAR instead of COCOLA | PHALAR · MoisesDB K=64 | Top-1 accuracy: 41.84 -> 70.87 | Improved |
| add spectral pooling and CVNN head to frozen MERT | PHALAR · MoisesDB K=64 | Top-1 accuracy: 41.84 (COCOLA) -> 45.85 | Improved |
| use semantic models (CLAP, CDPAM) for stem retrieval | PHALAR · MoisesDB K=64 | Top-1 accuracy: 1.24 and 1.15 | Hurt |
| use PHALAR score instead of FAD to rank generators | PHALAR · Stem generators (human ranking) | ranking agreement: 2.77 STAGE vs 2.55 SA-ControlNet | Improved |
| add label smoothing of 0.9 to InfoNCE positives | PHALAR · PHALAR training | stem retrieval | No clear effect |
Closest papers by meaning
- Contrastive Learning from Synthetic Audio Doppelgängers (ICLR 2025 · similarity 0.86)
- Supervised Contrastive Learning from Weakly-Labeled Audio Segments for Musical Version Matching (ICML 2025 · similarity 0.85)
- LLark: A Multimodal Instruction-Following Language Model for Music (ICML 2024 · similarity 0.85)
- CMI-RewardBench: Evaluating Music Reward Models with Compositional Multimodal Instruction (ICML 2026 · similarity 0.85)
- V2Meow: Meowing to the Visual Beat via Video-to-Music Generation (AAAI 2024 · similarity 0.84)
- MusicRL: Aligning Music Generation to Human Preferences (ICML 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
