Frequency-Temporal Attention Network for Singing Melody Extraction

TL;DR

This paper introduces a frequency-temporal attention network inspired by human auditory perception, designed to improve singing melody extraction from musical audio by focusing on frequency and temporal features.

Contribution

The paper presents a novel neural network architecture with frequency-temporal attention modules and a fusion mechanism, enhancing melody extraction accuracy over existing methods.

Findings

Outperforms state-of-the-art singing melody extraction methods

Effectively models frequency and temporal features of music

Achieves higher accuracy in melody detection

Abstract

Musical audio is generally composed of three physical properties: frequency, time and magnitude. Interestingly, human auditory periphery also provides neural codes for each of these dimensions to perceive music. Inspired by these intrinsic characteristics, a frequency-temporal attention network is proposed to mimic human auditory for singing melody extraction. In particular, the proposed model contains frequency-temporal attention modules and a selective fusion module corresponding to these three physical properties. The frequency attention module is used to select the same activation frequency bands as did in cochlear and the temporal attention module is responsible for analyzing temporal patterns. Finally, the selective fusion module is suggested to recalibrate magnitudes and fuse the raw information for prediction. In addition, we propose to use another branch to simultaneously predict the presence of singing voice melody. The experimental results show that the proposed model outperforms existing state-of-the-art methods.