Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Can pre-trained Deep Learning models predict groove ratings?

Axel Marmoret, Nicolas Farrugia, Jan Alexander Stupacher

Abstract

This study explores the extent to which deep learning models can predict groove and its related perceptual dimensions directly from audio signals. We critically examine the effectiveness of seven state-of-the-art deep learning models in predicting groove ratings and responses to groove-related queries through the extraction of audio embeddings. Additionally, we compare these predictions with traditional handcrafted audio features. To better understand the underlying mechanics, we extend this methodology to analyze predictions based on source-separated instruments, thereby isolating the contributions of individual musical elements. Our analysis reveals a clear separation of groove characteristics driven by the underlying musical style of the tracks (funk, pop, and rock). These findings indicate that deep audio representations can successfully encode complex, style-dependent groove components that traditional features often miss. Ultimately, this work highlights the capacity of advanced deep learning models to capture the multifaceted concept of groove, demonstrating the strong potential of representation learning to advance predictive Music Information Retrieval methodologies.

Can pre-trained Deep Learning models predict groove ratings?

Abstract

This study explores the extent to which deep learning models can predict groove and its related perceptual dimensions directly from audio signals. We critically examine the effectiveness of seven state-of-the-art deep learning models in predicting groove ratings and responses to groove-related queries through the extraction of audio embeddings. Additionally, we compare these predictions with traditional handcrafted audio features. To better understand the underlying mechanics, we extend this methodology to analyze predictions based on source-separated instruments, thereby isolating the contributions of individual musical elements. Our analysis reveals a clear separation of groove characteristics driven by the underlying musical style of the tracks (funk, pop, and rock). These findings indicate that deep audio representations can successfully encode complex, style-dependent groove components that traditional features often miss. Ultimately, this work highlights the capacity of advanced deep learning models to capture the multifaceted concept of groove, demonstrating the strong potential of representation learning to advance predictive Music Information Retrieval methodologies.

Paper Structure

This paper contains 18 sections, 2 equations, 3 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Data and Groove Ratings
  3. Audio Signal Processing
  4. Deep Audio Models
  5. Masked Acoustic Modeling
  6. CLAP & Text-Audio Contrastive Learning
  7. MIR features
  8. Predicting and Evaluating Groove
  9. Open-source statement
  10. Models Formalism
  11. Linear Probing - Ridge Regression
  12. Model Evaluation with $\mathop{\mathrm{R^{2}}}\limits$ score
  13. Analysis on Source-Separated Stems
  14. Visualizations
  15. Results and Discussion
  16. ...and 3 more sections

Figures (3)

  • Figure 1: Repartition of the groove ratings for all 148 songs. For each song, results are averaged across all participants.
  • Figure 2: Scatter plots of the predicted values against the ground truth values, for all ratings, and for the MuQ embeddings.
  • Figure 3: Visualization of the projection on the first two principal components (PCA) of audio embeddings from the MuQ model derived from source-separated instrumental stems