Objective analysis of the singing voice as a training aid

Area of Science:

• Acoustics
• Music Technology
• Speech Science

Background:

• Accurate analysis of singing voice parameters is crucial for vocal training and performance.
• Existing methods for tracking fundamental frequency and vocal characteristics have limitations, especially with complex vocalizations.

Purpose of the Study:

• To introduce a novel tool for robust tracking of fundamental frequency (F0) in singing.
• To develop an objective measure for key singing voice parameters, including vibrato rate, vibrato extent, and vocal intonation.
• To evaluate the performance of the proposed techniques against classical Fourier-based approaches.

Main Methods:

• Implementation of high-resolution Power Spectral Density (PSD) estimation.
• Utilization of AutoRegressive (AR) models for reliable formant tracking.
• Application of the developed techniques to a dataset of approximately 1000 vocalizations from diverse singers.

Main Results:

• The proposed tool demonstrates superior performance in tracking fundamental frequency and vocal parameters compared to traditional Fourier-based methods.
• Reliable formant tracking is achieved even in vocalizations with highly variable characteristics.
• The techniques show effectiveness across vocalizations from both professional and non-professional singers.

Conclusions:

• The developed tool provides a robust and objective method for analyzing singing voice parameters.
• Real-time analysis capabilities could significantly benefit non-professional singers and vocal instructors.
• The tool offers reliable measures for evaluating vocal improvement during and after training.