Key detection through pitch class distribution model and ANN

  • Authors:

  • Jiayin Sun;Haifeng Li;Li Lei

  • Affiliations:

  • School of Computer Science and Technology, Harbin Institute of Technology, Harbin, China;School of Computer Science and Technology, Harbin Institute of Technology, Harbin, China;School of Computer Science and Technology, Harbin Institute of Technology, Harbin, China

  • Venue:
  • DSP'09 Proceedings of the 16th international conference on Digital Signal Processing
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

Being one of the most well-known high-level music content concepts, music key reveals an important theoretical feature for Western music in music structure analysis, music note and chord transcription and music mood comprehension. Key finding becomes a main task of Music Information Retrieval (MIR). In this paper, we propose a novel approach with good robustness to detect keys in polyphonic music based on Artificial Neural Network (ANN). Constant Q transform (CQT) is firstly applied to music signal for CQT spectrum analysis. Then onset detection and pitch tuning are introduced in order to ensure a high robustness. Finally a distribution matrix is generated as music key feature. Considering the classifier, a neural network is applied to model the pitch class distribution and complete the task of key recognition. Experiments showed that the proposed strategy can reach a good performance in polyphonic music at a relatively lower computational cost, and proved our strategy to be quite promising.