In the process of music education, extracting timbre characteristics from a singer’s voice is an indispensable step for grasping musical style and enhancing vocal performance skills. This paper first proposes a method for extracting timbre characteristics in vocal techniques for music education. It employs autocorrelation and MFCC methods to extract fundamental frequency for modeling musical waveform spectra. Subsequently, the proposed timbre extraction method is validated through experiments analyzing vocal timbre characteristics and error analysis. Furthermore, a scoring method for evaluating singers’ performance levels is presented and experimentally tested. Results demonstrate that the proposed evaluation criteria exhibit the highest consistency with expert assessments, achieving correlation coefficients of 0.7826 and 0.7687. Both values surpass existing pitch/rhythm evaluation metrics, thereby validating the effectiveness of this methodology.
