Quantifying auditory temporal stability in a large database of recorded music

Area of Science:

• Music Information Retrieval
• Computational Auditory Processing
• Human-Computer Interaction

Background:

• Rhythmic interaction with music is fundamental to human experience.
• Automated beat detection in music has applications in exercise, rehabilitation, and entertainment.
• Existing algorithms lack quantification of beat temporal stability.

Purpose of the Study:

• To propose a novel method for quantifying the temporal stability of musical beats.
• To introduce the Balanced Evaluation of Auditory Temporal Stability (BEATS) algorithm.
• To facilitate personalized music playlist generation for specific applications.

Main Methods:

• Development of the Balanced Evaluation of Auditory Temporal Stability (BEATS) algorithm.
• Application of BEATS to the Million Song Dataset for analysis.
• Creation of a web interface combining BEATS statistics with music metadata.

Main Results:

• BEATS successfully quantifies the temporal stability of detected beats in music.
• The Million Song Dataset was utilized to demonstrate the algorithm's effectiveness.
• A user-friendly web interface was developed for accessing BEATS statistics and metadata.

Conclusions:

• BEATS offers a new metric for analyzing musical temporal structure.
• The developed system supports research into music perception and personalized playlist creation.
• Applications include clinical rehabilitation, fitness training, and enhanced music enjoyment.