Speech sound acquisition, coarticulation, and rate effects in a neural network model of speech production.

Abstract

This article describes a neural network model of speech motor skill acquisition and speech production that explains a wide range of data on variability, motor equivalence, coarticulation, and rate effects. Model parameters are learned during a babbling phase. To explain how infants learn language-specific variability limits, speech sound targets take the form of convex regions, rather than points, in orosensory coordinates. Reducing target size for better accuracy during slower speech leads to differential effects for vowels and consonants, as seen in experiments previously used as evidence for separate control processes for the 2 sound types. Anticipatory coarticulation arises when targets are reduced in size on the basis of context; this generalizes the well-known look-ahead model of coarticulation. Computer simulations verify the model's properties.The primary goal of the modeling work described in this article is to provide a coherent theoretical framework that provides explanations for a wide range of data concerning the articulator movements used by humans to produce speech sounds. This is carried out by formulating a model that transforms strings of phonemes into continuous articulator movements for producing these phonemes. This study of speech production is largely motivated by the following question of speech acquisition: How does an infant acquire the motor skills needed to produce the speech sounds of his or her native language? Speech production involves complex interactions among several different reference frames. A phonetic frame describes the sounds a speaker wishes to produce, and the signals that convey these sound units to a listener exist within an acoustic frame. Tactile and proprioceptive signals form an orosensory frame (e.g., Perkell, 1980) that describes the shape of the vocal tract, and the muscles controlling the positions of individual articulators make up an articulatory frame. The parameters governing the interactions among these frames cannot be fixed at birth. One reason for this is the language specificity of these interactions. For example, English listeners distinguish between the sounds /r/ and /!/, but Japanese listeners do not. Corresponding differences are seen in the articulator movements of the two groups (Miyawaki et al., 1975). Thus, despite some obvious commonalities between the phonetics of different languages (e.g., widespread use of consonants like /d/, /n/, and /s/ across the world's languages), the precise nature of mappings between acoustic goals and articulator movements depends on the lanThis research was supported in part by Air Force Office of Scientific Research F49620-92-J-0499. I would like to thank Dan Bullock and Elliot Saltzman for their insightful suggestions on an earlier

A vector-integration-to-endpoint model for performance of viapoint movements

“…Two examples will suffice to indicate some of the issues. First, it is well known in speech production that targets are better treated as regions than as points, and Guenther ( 1995) recently incorporated ideas from the VITE and DIRECT models in his DIY A (Directions Into Velocities of Articulators) model, which incorporates such regions as a basic postulate. It may be that the emergence of such regions can be attributed to the use of learned TTC thresholds to terrrlinate approaches to points near the center of such regions.…”

“…They also offered an explanation for kinematic properties of reaching movements that included the bell shaped velocity profile, Woodworth's law, Fitts's law and variations in velocity profile symmetry as a function of movement duration. The model has been extended in the form of the DIRECT and DIY A models to incorporate inverse differential transformations between task space and motor coordinates for both reaching Fiala, 1994Fiala, , 1995 and speech production (Guenther, 1992;Guenther, 1994Guenther, , 1995. These transformations were shown to enable motor equivalence and to be learnable with the help of perceptual information generated by action.…”

A theoretical investigation of reference frames for the planning of speech movements.

“…Such a mapping has been termed a "forward model" by Jordan (1990) and has been used in different capacities in adaptive models of speech production (e.g., Bailly eta!., 1991; Guenther, 1994Guenther, , 1995a and other motor tasks such as reaching (e.g., Bullock, Grossberg, and Guenther, 1993;Jordan, 1990). A typical neural network construct for learning a forward model is illustrated in Figure 3.…”

“…This replaces the constriction-based planning frame used in earlier versions ofthe DIVA model (Guenther, 1994(Guenther, , 1995a. The Planning Position Vector in the model represents the current state of the vocal tract within the auditory perceptual reference frame.…”

Cortical interactions underlying the production of speech sounds

“…The model is implemented in computer simulations that control an articulatory synthesizer (Maeda, 1990) in order to produce an acoustic signal. The articulator movements and acoustic signal produced by the model can be compared to the productions of human speakers; the results of many such comparisons are described elsewhere (Callan, Kent, Guenther, & Vorperian, 2000;Guenther, 1995;Guenther, Hampson, & Johnson, 1998;Guenther et al, 1999;Nieto-Castanon, Guenther, Perkell, & Curtin, 2005;Perkell et al, 2004a,b).…”