Abstract:  New power spectral density level of Chinese speech was obtained in this paper by using the method of speech noise, a method found to be superior in many respects.The speech noise was the phenomena produced by
many people speaking together. Based on existing methods for measuring and calculating the average spectrum of: language，this paper adopted some improvements in the following respects; measurement methods，mcasurcmcm
system, experiment envmonment，speech materials, experiment subjects and calculation methods. In this paper, every Chinese speech material selected typically reflects characteristics of Chinese speech. We chose 32 subjects(16
male and 16 female) for the experiment，each of them has obtained Putonghua Proficiency Rating Certificate, and their ages arc in the range from 20一30. Every subject’s Chinese speech samples were recorded by multichanncl
measurement system, and finally 160 Chinese speech samples were obtained. The experiment was conducted in the anechoic room of Nanjing University to improve the signal to noise ratio. In the application range of the anechoic
room we used，the anechoic room is basically free field. Before each experiment，the instruments were calibrated. Every subjects long-term spectrum of the sound pressure was measured by the multi-functional software using fast
Fouricr transform method with 1/3 octave. Subjective evaluation and objective evaluation were used for excluding abnormal data, 5 samples were excluded finally. According to the 155 valid speech samples, the average spectrum of
Chinese speech was achieved by the calculation process mentioned below; normalized every subjects lon}term sound pressure spectrum, mixed all subjects’ normalized sound pressure spectrum, normalized the mixed samples as the
long-term average overall sound pressure level is 0 dB, and finally calculated the power spectral density level of Chinese speech. The result showed the main energy of Chinese speech spreading in the frequency range of 63一20
kHz. The average spectrum of Chinese speech provides the basis for the design of clectro-acoustic device and all the systems of generating, transmitting, receiving and processing Chinese speech signals.