Hideyuki Sawada

Gesture recognition using an acceleration sensor and its application to musical performance control

It is understood that nonverbal means of communication are more important than verbal communication in the transmission of intention or emotion. Gestures using body or hand are typical nonverbal means of communication. Human intention or emotion appears to be expressed to a greater extent by the force applied to the body than by the position of the hand. The authors noted that the force acting during motion can be sensed as an acceleration and therefore attempted recognition of gestures with a three-dimensional acceleration sensor. In the experiments using this system, the changes of the acceleration, the rotational force, and the directional distribution of the acceleration are determined from the sequential three-dimensional acceleration data, as features of the motion. By examining the extent of matching to standard patterns, ten kinds of gestures can be discriminated with a nearly 100 percent recognition rate. A real-time music control system is also constructed by applying the proposed method. Compared to the conventional methods using image processing or the data gloves, in the new system the delay in tempo detection is shorter and the mechanism is simpler. The system gives a clue to the design of flexible and sensitive man-machine interfaces based on gesture.

Singing performance of the talking robot with newly redesigned artificial vocal cords

The authors are developing a talking robot which is a mechanical vocalization system modeling the human articulatory system. The talking robot is constructed with mechanical parts that are made by referring to human vocal organs biologically and functionally. In this study, a newly redesigned artificial vocal cords is developed for the purpose of extending the speaking capability of the talking robot. The artificial vocal cords is developed using functional approach. A thin layer of rubber band with a width of 7mm is attached on a plastic body in a sealed chamber creates an artificial sound source. The fundamental frequency of the sound source vary from 80Hz to 250 Hz depending on the pressure to the rubber band and the curving shape of the rubber band. The curving shape of the rubber band is adjustable by an innovative design mechanism driven by a servo motor. The amount of air pressure applied to the rubber band is controlled by another servo motor. The experiments to verify the speak capability of the talking robot with this newly redesigned vocal cords is conducted by letting the robot generating five vowel sounds with different combinations of 20 levels of rubber band shape and 5 levels of air pressure input. The pitch of each sound is extracted to determine the effect of the rubber band shape and the pressure input on the output sound. The result shows that this newly redesigned vocal cords greatly increase the speaking capability of the talking robot, especially its singing performance.

Efficacy of a new microvibration sensation measurement device at detecting diabetic peripheral neuropathy using a newly devised finger method

To investigate the efficacy of the finger method using a new microvibration sensation measurement device in the evaluation of diabetic peripheral neuropathy (DPN). A cross‐sectional study of 52 type 2 diabetic outpatients was performed. Patients were evaluated for DPN using American Diabetes Association (ADA) criteria, Michigan Neuropathy Screening Instrument, and the finger method. Patients were classified into probable DPN or non‐DPN groups, according to ADA criteria. The finger method measured peripheral neuropathy vibration (PNV) score of index and middle fingers using the new device in three procedures: PNV 1, PNV 4, and PNV 8. PNV scores ranged from 1 to 30 and were compared between the two groups. The PNV scores were significantly higher in the DPN group (P < .01). The PNV scores for right fingers of DPN and non‐DPN groups were 10.2 ± 7.4 and 3.4 ± 3.3 by PNV 1, 20 ± 4.9 and 10.7 ± 5.3 by PNV 4, and 23.2 ± 4.9 and 14.6 ± 7.8 by PNV 8. Our data suggest that the finger method performed with the new device is useful in the evaluation of DPN.

Sound generation using twiddle interface

Although algorithmic sound syntheses, such as a granular synthesis and a physical model based synthesis, have been studied actively and have come into practical use, the development of a human-machine interface is not sufficient for the translation of emotional feelings into sound. This paper presents two sound generation systems in which sounds are generated by the direct manipulations of interface devices. A pair of datagloves and a specially designed equipment which measures grasping forces are employed in the systems. The approach is considered to provide a new sound generation techniques in the music scene together with the flexible man-machine interface.

Simplified cerebellum-like spiking neural network as short-range timing function for the talking robot

ABSTRACT In human speech, the timing function is important for determining its duration, stress and rhythm; however, little attention has been paid to these issues when building a speech synthesis system. In the human brain, the cerebellum plays a key role in the coordination, precision and timing of motor responses. We have developed a talking robot, which generates human-like vocal sounds using a simplified cerebellum-like neural network model as the timing function. The model was designed using the System Generator software in Matlab environment and the timing duration of trained speech was estimated using hardware co-simulated with a field programmable gate array board (FPGA). The timing information obtained from the co-simulation, together with the output motor vector, is sent to the talking robot controller in order to generate vowels of short, medium and long duration. Using this model for short-range timing of less than 1200 milliseconds, we verify that the short-range learning capability of the cerebellar-like neural network is applicable to the speaking robot for generating a human-like speech with prosodic features.

Text-to-speech of a talking robot for interactive speech training of hearing impaired

The authors are developing a talking robot which is a mechanical vocalization system modeling the human articulatory system. The talking robot is constructed with mechanical parts that are made by referring to human vocal organs biologically and functionally. In this study, a newly redesign artificial vocal cord is developed for the purpose of extending the speaking capability of the talking robot. The artificial vocal cord is developed using functional approach. A thin layer of rubber band with a width of 7mm is attached on a plastic body in a sealed chamber creates an artificial sound source. The fundamental frequency of the sound source vary from 80Hz to 280 Hz depending on the pressure to the rubber band and the curving shape of the rubber band. The curving shape of the rubber band is adjustable by an innovative design mechanism driven by a servo motor. The amount of air pressure applied to the rubber band is controlled by another servo motor which is referred as volume motor in this study. The experiment to verify the speak capability of the talking robot with this newly redesigned vocal cord is conducted by letting the robot generating five vowel sounds with difference combination of 20 levels of rubber band shape and 10 levels of air pressure input. The pitch and volume of each sound is extracted to determine the effect of the rubber band shape and the pressure input on the output sound. The result shows that this newly redesigned vocal cord greatly increase the speaking capability of the talking robot, especially its singing performance.

Cerebellum-like neural network for short-range timing function of a robotic speaking system

The timing control is necessary for determining its duration, stress, and rhythm in human speech; however, little attention has been paid to these issues when building a speech synthesis system. We have developed a talking robot, which generates human-like vocal sounds. The cerebellum is an important part of human brain organ that has a significant role in the coordination, precision, and timing of motor responses. In this study, we develop a simplified cerebellumlike spiking neural network model to control the timing function for the talking robot. The model was designed using the System Generator software in Matlab, and the timing duration of trained speech was estimated using hardware cosimulated with a field programmable gate array board (FPGA). The timing information obtained from the co-simulation, together with the output motor vector, is sent to the talking robot controller to generate a sound with a short duration. The result indicates that this model can be used for short-range timing learning of the talking robot.