A Review on Feature Extraction Using CNN in Musical Instrument Recognition

A Review on Feature Extraction Using CNN in Musical Instrument Recognition

Dr. Seema R. Chaudhary1, Dr. Supriya A. Kinariwala2

1Department of Computer Science & Engineering, MIT, Chh.Sambhajinagar, Maharashtra (India)

2Department of Emerging Sci. & Technology, MIT, Chh.Sambhajinagar, Maharashtra (India)

---------------------------------------------------------------------***---------------------------------------------------------------------

Abstract - This literature review explores the use of convolutional neural networks (CNNs) for the classification of musical instruments within music information retrieval. It highlights the limitations of traditional manual feature extraction methods and emphasizes the advantages of deep learning in automating feature extraction from raw audio data. By focusing on time-frequency representations, CNNs have demonstrated significant effectiveness in identifying relevant features crucial for tasks such as music genre identification, transcription, and music recommendation systems.

Key Words:  feature extraction, CNN, MIR, music instrument recognition