Investigating the role of musical genre in human perception of music stretching resistance

Jun Chen, Chaokun WangarXiv 2017 · arXiv

Introduction

In modern audio production, time-stretching—the process of changing the tempo of a recording without altering its pitch—is a fundamental tool. Whether syncing a soundtrack to video or adjusting a song's length for animation, the ability to manipulate duration is essential. However, stretching is not a "free" operation. Beyond certain limits, audio develops perceptual artifacts or simply feels "wrong" to the listener. This paper explores Music Stretching Resistance (MSR), defined as the limit to how much a song can be sped up or slowed down before it becomes perceptually unacceptable.

While previous computational models have attempted to estimate these limits, the underlying factors influencing human tolerance remained largely unexplored. This research demonstrates that a song's genre is the primary predictor of its stretchability, far outweighing technical features like pitch or timbre.

Methodology

The researchers conducted a month-long user study involving 17 participants, primarily non-musicians, to evaluate a dataset of 894 songs across 11 distinct genres.

Technical Setup and Annotation

The study utilized the Synchronized Overlap-Add (SOLA) algorithm to modify time scales while preserving pitch. Each song was rendered into 98 versions, ranging from 0.02x to 1.98x the original speed. Participants identified two critical thresholds: $\alpha_{min}$ (the fastest acceptable compression) and $\alpha_{max}$ (the slowest acceptable elongation).

The dataset included genres such as Pop, Rock, Easy Listening, Folk, Latin, Country, Hip-hop & Rap, R&B, Jazz & Blues, Classical, and Electronic. Genres were verified using metadata from Wikipedia and popular music platforms.

The MSR Rectangle

To quantify and visualize the relationship between genres, the authors introduced the "MSR Rectangle." This coordinate-based method plots $\alpha_{min}$ on the x-axis and $\alpha_{max}$ on the y-axis. By calculating the overlap (Jaccard similarity) between these rectangles, the researchers could mathematically define the similarity between musical styles based solely on their stretching resistance.

Figure 6: Visualization of the MSR Rectangle method, illustrating regional analysis and genre overlap.

Results and Discussion

Analysis of Variance (ANOVA) results were highly significant ($P < 0.001$), confirming that genre is a dominant factor in shaping MSR. The study revealed several key insights:

Genre-Specific Elasticity

MSR is remarkably consistent within a genre ("intra-genre cohesiveness") but varies significantly between them ("inter-genre discrepancy").

  • High Flexibility: Classical and Easy Listening genres proved highly "stretchable" ($\alpha_{min} \approx 0.58$, $\alpha_{max} \approx 1.38$). These styles often lack a rigid, driving beat or dense lyrical content, allowing for wider variations in tempo.
  • High Rigidity: Hip-hop, Rap, and R&B exhibited the narrowest stretching ranges ($\alpha_{min} \approx 0.79$, $\alpha_{max} \approx 1.21$). The authors suggest this is due to their reliance on a "solid tempo" and dense lyrics; excessive stretching quickly makes lyrics unintelligible or the rhythm unnatural.
Figure 1: Average MSR thresholds (alpha_min and alpha_max) across different musical genres.

Perceptual Symmetry and Tempo Independence

Interestingly, human tolerance for stretching is generally symmetric: if a genre can be compressed by 20%, it can typically be elongated by approximately 20%. Furthermore, the original tempo (BPM) of a song showed almost no correlation with its MSR. A fast song and a slow song within the same genre generally share the same percentage-based stretching limits.

Conclusion

This research establishes genre as a critical factor in the human perception of time-stretched music. The introduction of the MSR Rectangle provides a robust framework for quantifying these perceptual limits. While the study was limited by a small participant demographic and the use of a specific algorithm (SOLA), the findings have significant implications for "content-aware" music adaptation. Future work may explore how more advanced AI-based stretching algorithms expand these "safe" zones and how these insights can be applied to live performance technology.

Summary updated

Abstract

To stretch a music piece to a given length is a common demand in people's daily lives, e.g., in audio-video synchronization and animation production. However, it is not always guaranteed that the stretched music piece is acceptable for general audience since music stretching suffers from people's perceptual artefacts. Over-stretching a music piece will make it uncomfortable for human psychoacoustic hearing. The research on music stretching resistance attempts to estimate the maximum stretchability of music pieces to further avoid over-stretch. It has been observed that musical genres can significantly improve the accuracy of automatic estimation of music stretching resistance, but how musical genres are related to music stretching resistance has never been explained or studied in detail in the literature. In this paper, the characteristics of music stretching resistance are compared across different musical genres. It is found that music stretching resistance has strong intra-genre cohesiveness and inter-genre discrepancies in the experiments. Moreover, the ambiguity and the symmetry of music stretching resistance are also observed in the experimental analysis. These findings lead to a new measurement on the similarity between different musical genres based on their music stretching resistance. In addition, the analysis of variance (ANOVA) also supports the findings in this paper by verifying the significance of musical genre in shaping music stretching resistance.
BibTeX
@misc{chen2017investigating,
  author = {Jun Chen and Chaokun Wang},
  title = {{Investigating the role of musical genre in human perception of music stretching resistance}},
  year = {2017},
  eprint = {1701.03274},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1701.03274},
}
Figure