Long-term average spectral characteristics of different Cantonese opera singing styles

Abstract

Cantonese Opera is a valuable cultural heritage populates in China. Basic singing styles consist of zi hou, ping hou and da hou. However, objective parameters measuring voice qualities in Cantonese opera singing are lacking. The current study examined the sound quality associated with zi hou, ping hou and da hou singing styles in comparison to conversational voice by means of Long-Term Averaged Spectra (LTAS). Continuous singing and speech samples were obtained from professional Cantonese opera singers and na?ve speakers of Cantonese. All singing and speech samples were digitized at 44 kHz and 16 bits/ sample. Parameters including the first spectral peak (FSP), mean spectral energy (MSE), spectral tilt (ST) and high frequency energy (HFE) were derived from the LTAS contours by using Praat. Different singing styles exhibited different LTAS contours and were associated with significantly a higher ST value than conversational voice, implying a difference in resonance. Further investigation on the phonatory mechanism is indicated.