CRNN-Based Hierarchical Musician Identification in Classical Music

View at publisher Summarise with VIRAT
Authors: Sabrina Jamil, Harshini Ganapathi, Navya Malhotra, Sasan Haghani
Cited by: 0
References: 8

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

This paper presents a machine-learning framework for identifying classical composers and performers based on stylistic characteristics embedded in audio recordings. Unlike conventional music recognition systems that rely on audio fingerprinting, the proposed approach focuses on expressive musical features such as articulation, phrasing, dynamics, and temporal shaping—properties that encapsulate interpreters’ unique performance styles. The system integrates mel-spectrograms and Mel-Frequency Cepstral Coefficients (MFCCs) with a Convolutional Recurrent Neural Network (CRNN) architecture that captures both spatial and temporal attributes of musical expression. A hierarchical classification pipeline first determines the composer and subsequently refines the prediction to the performer within each composer group. The framework is supported by a user interface that accepts audio recordings and returns the model’s predictions in real time. Our system integrates MFCC-based feature extraction with a hierarchical CRNN that performs both composer and performer classification. By modeling expressive cues and interpretive signatures, it achieves performer-level distinction, a capability largely absent in current music information retrieval (MIR) systems.

Fields of Study & Subjects

Music and Audio Processing Music Technology and Sound Studies Musicology and Musical Analysis

Publication details

Journal / source
International Conference on Consumer Electronics
Pages
1–5
Publication date
3 February 2026
Language
EN