Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Who Will Top the Charts? Multimodal Music Popularity Prediction via Adaptive Fusion of Modality Experts and Temporal Engagement Modeling

Yash Choudhary, Preeti Rao, Pushpak Bhattacharyya

TL;DR

This work tackles pre-release music popularity prediction by introducing Career Trajectory Dynamics (CTD) features and GAMENet, a gated adaptive multimodal network that fuses audio, lyrics, and social signals. CTD captures aggregate and temporal artist and song engagement, yielding large gains, while GAMENet's adaptive fusion improves predictions over simple feature concatenation, achieving strong R^2 on both Music4All-CTD and SPD_Cleaned datasets. The results highlight the dominant role of social engagement signals, with lyrics and audio providing complementary context, and demonstrate robustness across datasets. Overall, the approach advances interpretable, temporally aware multimodal modeling for practical strategic decisions in music marketing.

Abstract

Predicting a song's commercial success prior to its release remains an open and critical research challenge for the music industry. Early prediction of music popularity informs strategic decisions, creative planning, and marketing. Existing methods suffer from four limitations:(i) temporal dynamics in audio and lyrics are averaged away; (ii) lyrics are represented as a bag of words, disregarding compositional structure and affective semantics; (iii) artist- and song-level historical performance is ignored; and (iv) multimodal fusion approaches rely on simple feature concatenation, resulting in poorly aligned shared representations. To address these limitations, we introduce GAMENet, an end-to-end multimodal deep learning architecture for music popularity prediction. GAMENet integrates modality-specific experts for audio, lyrics, and social metadata through an adaptive gating mechanism. We use audio features from Music4AllOnion processed via OnionEnsembleAENet, a network of autoencoders designed for robust feature extraction; lyric embeddings derived through a large language model pipeline; and newly introduced Career Trajectory Dynamics (CTD) features that capture multi-year artist career momentum and song-level trajectory statistics. Using the Music4All dataset (113k tracks), previously explored in MIR tasks but not popularity prediction, GAMENet achieves a 12% improvement in R^2 over direct multimodal feature concatenation. Spotify audio descriptors alone yield an R^2 of 0.13. Integrating aggregate CTD features increases this to 0.69, with an additional 7% gain from temporal CTD features. We further validate robustness using the SpotGenTrack Popularity Dataset (100k tracks), achieving a 16% improvement over the previous baseline. Extensive ablations confirm the model's effectiveness and the distinct contribution of each modality.

Who Will Top the Charts? Multimodal Music Popularity Prediction via Adaptive Fusion of Modality Experts and Temporal Engagement Modeling

TL;DR

This work tackles pre-release music popularity prediction by introducing Career Trajectory Dynamics (CTD) features and GAMENet, a gated adaptive multimodal network that fuses audio, lyrics, and social signals. CTD captures aggregate and temporal artist and song engagement, yielding large gains, while GAMENet's adaptive fusion improves predictions over simple feature concatenation, achieving strong R^2 on both Music4All-CTD and SPD_Cleaned datasets. The results highlight the dominant role of social engagement signals, with lyrics and audio providing complementary context, and demonstrate robustness across datasets. Overall, the approach advances interpretable, temporally aware multimodal modeling for practical strategic decisions in music marketing.

Abstract

Predicting a song's commercial success prior to its release remains an open and critical research challenge for the music industry. Early prediction of music popularity informs strategic decisions, creative planning, and marketing. Existing methods suffer from four limitations:(i) temporal dynamics in audio and lyrics are averaged away; (ii) lyrics are represented as a bag of words, disregarding compositional structure and affective semantics; (iii) artist- and song-level historical performance is ignored; and (iv) multimodal fusion approaches rely on simple feature concatenation, resulting in poorly aligned shared representations. To address these limitations, we introduce GAMENet, an end-to-end multimodal deep learning architecture for music popularity prediction. GAMENet integrates modality-specific experts for audio, lyrics, and social metadata through an adaptive gating mechanism. We use audio features from Music4AllOnion processed via OnionEnsembleAENet, a network of autoencoders designed for robust feature extraction; lyric embeddings derived through a large language model pipeline; and newly introduced Career Trajectory Dynamics (CTD) features that capture multi-year artist career momentum and song-level trajectory statistics. Using the Music4All dataset (113k tracks), previously explored in MIR tasks but not popularity prediction, GAMENet achieves a 12% improvement in R^2 over direct multimodal feature concatenation. Spotify audio descriptors alone yield an R^2 of 0.13. Integrating aggregate CTD features increases this to 0.69, with an additional 7% gain from temporal CTD features. We further validate robustness using the SpotGenTrack Popularity Dataset (100k tracks), achieving a 16% improvement over the previous baseline. Extensive ablations confirm the model's effectiveness and the distinct contribution of each modality.

Paper Structure

This paper contains 16 sections, 4 equations, 4 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Dataset
  4. Music4All Family
  5. SpotGenTrack Popularity Dataset
  6. Dataset Cleaning and Pre-processing
  7. Methodology
  8. Career Trajectory Dynamics (CTD) Features
  9. OnionEnsembleAENet
  10. GAMENet: Gated Adaptive Modality Experts Network
  11. Experiments and Results
  12. Experimental Setup
  13. Ablation Study: Impact of CTD Features
  14. MultiModal Baseline on Music4All
  15. GAMENet Results
  16. ...and 1 more sections

Figures (4)

  • Figure 1: Distribution of Spotify popularity scores in the original Music4All dataset.
  • Figure 2: Distribution of release years in the Music4All dataset, showing a sharp concentration in recent decades with mean 2006 and median 2011
  • Figure 3: Cleaning and processing pipeline from Music4All and Music4All-ONION to the final M4A-CTD dataset containing 53,777 tracks.
  • Figure 4: GAMEnet