Music identification system using MPEG-7 audio signature descriptors.

Area of Science:

• Digital Signal Processing
• Music Information Retrieval
• Copyright Law

Background:

• The proliferation of digital music has led to widespread copyright infringement.
• Existing music identification systems often struggle with speed and accuracy.
• The need for robust systems to detect illegally copied music is critical.

Purpose of the Study:

• To develop a multiresolution music identification system for detecting copyright infringement.
• To enhance both the speed and accuracy of music identification processes.
• To propose methods for determining decision thresholds in music identification.

Main Methods:

• Utilizing MPEG-7 audio signature descriptors for music identification.
• Implementing a multiresolution approach with low-resolution descriptors for candidate searching.
• Employing full-resolution descriptors for precise identification of query audio.
• Developing two distinct methods for setting decision thresholds.

Main Results:

• The proposed system achieves high speed and accuracy in music identification.
• Method II for decision threshold determination demonstrated a 99.4% accuracy rate.
• The system proved effective for query audio both within and outside the database.
• Simulation results validate the system's performance.

Conclusions:

• The developed multiresolution system is highly effective for music identification.
• The proposed system shows significant potential for copyright control applications.
• The high accuracy and speed make it suitable for real-world deployment.