System resolution in perception and production: Implications for speech signal processing and speech motor control
C. K. Chuang · The Journal of the Acoustical Society of America · 1979
One global yet rather basic question in regard to speech is the extent to which underlying organic capabilities are utilized for speech functions. Applying the concept of information processing uncertainty to perceptual stimulus discriminability and the productional variability, a unified definition of the system resolution for both processes was derived. The resulting perceptual system resolution for pitch, intensity, duration, and average formant frequency is 4.8 bits, 3.4 bits, 3.6 bits and 3.8 bits, respectively, and the corresponding resolution for production in those dimensions is 4.6 bits, 3.3 bits, 3.2 bits and 3.6 bits, respectively. These results show that the system resolution in the speech mechanism all cluster between 3 and 5 bits. Namely the use of organic resolving function for speech processes do not exceed 32 decimal categories. Surprisingly the perceptual and production function differences are less than 0.5 bits in any dimension. This comparison reveals a very close system resolution match between production and perceptual processes. Implications of this observation in relation to the nonlinear resolution of vowel production and perception will be discussed. [Work supported by NS 13274.]