Sound power of speech between 63Hz and 20kHz at a normal vocal effort level

This paper reports anechoic measurements of speech (720 IEEE sentences) produced at a normal vocal effort by 12 native British English speakers (six male, six female) over a hemispherical region (1m radius) on one side of the talker to allow calculation of the sound power as well as an assessment of the on-axis sound pressure level. One-third octave bands were measured from 63Hz to 20kHz to include the Extended High Frequency (EHF) range above 7kHz. In the low-frequency range (below 200Hz), male talkers had higher levels than female talkers in the 100Hz and 125Hz one-third octave bands, and only male talkers had measurable levels in the 63Hz and 80Hz bands. In the mid-frequency range (200Hz – 4kHz) there was a difference between male and female talkers in the 800Hz one-third octave band where females were ≈ 3dB higher than males for the on-axis sound pressure level and ≈ 2dB higher for the sound power level. In the high-frequency range (5kHz – 20kHz), no statistically significant differences were found between male and female talkers; although there were large variations between talkers and between sentences that did, and did not contain the fricatives /s, z, S, Z, f, v, T, D/ and the unvoiced affricate /tS/. The results show that the definition of a unisex talker spectrum (average of male and female talkers) for the on-axis sound pressure level or sound power level is reasonable between the 160Hz and 20kHz one-third octave bands, and between the 125Hz and 16kHz octave bands.