Expiration Training System for Treatment of People with Speech Problems

Graduate School of Engineering, University of Fukui, 3-9-1 Bunkyo, Fukui 910-8507, Japan 1Career Design Laboratory for Gender Equality, Kanazawa University, Kakuma-machi, Kanazawa 920-1192, Japan 2Fukui College of Health Sciences, 55-13-1 Egamicho, Fukui 910-3190, Japan 3The National Institute of Vocational Rehabilitation, 3-1-3 Wakaba, Mihama-ku, Chiba 261-0014, Japan 4Hyogo Children’s Sleep and Development Medical Research Center, 1070 Akebono-cho, Nishi-ku, Kobe 651-2181, Japan 5Department of Child and Adolescent Psychological Medicine, University of Fukui Hospital, Matsuoka-shimoaizuki, Eiheiji-cho, Yoshida-gun, Fukui 910-1193, Japan 6Fukui Prefectural Rehabilitation Center for Children with Disabilities, 2-8-1 Yotsui, Fukui 910-0846, Japan 7Faculty of Education and Regional Studies, University of Fukui, 3-9-1 Bunkyo, Fukui 910-8507, Japan


Introduction
Various training methods have been implemented by speech therapists for people with speech, language and hearing disorders. (1)(2)(3)(4) Among them, expiration training is considered highly important for improving motor functions of articulatory organs. One expiration training method is the "blowing technique", in which a trainee blows air through a straw placed at a depth of 5-10 cm in a cup of water. The expiration power necessary for making an utterance (5-10 cm H 2 O) and duration (5 s) are verified, and then the strength of respiratory muscles is increased accordingly. (5,6) Various pronunciation techniques giving visual feedback of breath control have been conducted for pronunciation training. These include blowing at a candle (rhythmically, alternating between strong and weak blows, so as not to put out the candle) and blowing feather flutes, party horns, paper balloons, blowing balls, ping-pong balls, and pinwheels, among others. (7) For vocalizing continuous sounds (smoothly shifting from one single sound to another), blowing that can be visually confirmed is considered effective (8) because continuous long expirations are required.
We have developed a system in which trainees can imitate therapists' expiration patterns (e.g., timing, strength, and length of expiration) by visually examining the patterns in a graph shown through a computer. In this method, a trainee holds a mouthpiece with a chosen shape and simultaneously watches a graph indicating his/her expiration patterns. (9,10) It was confirmed that training using this system was effective for decreasing the differences in expiration patterns between a therapist and a trainee, which were calculated from the graph.

Structure of system
The structure of the system is shown in Fig. 1, expiration training reported in this paper was conducted with this system. The system includes various elements such  as CCD camera input, voice output, and mike input, among others, and can deal with various symptoms of people with speech, language and hearing disorders (i.e., trainees).

Measurement of volume of expiratory flow
The device for measuring the expiratory flow rate is shown in Fig. 2. The device consists of the following sensor units and an A/D converter. The data from the unit is fed into a computer.
• Air flow sensor 30-1000 (ml/s): FD-A1, Keyence • Amplifier unit: FD-V40A, Keyence • A/D converter 16 (bit)・1MSPS: CSI-320416, Interface In this system, the sampling period is set as 10 (ms) and the resolution of the A/D converter is set at 12 (bit). Figure 3 is a graph showing the expiration data of a healthy male adult while blowing rhythmically as measured using the air flow sensor.

Example of expiration training
The computer menu used in the expiration training system is shown in Fig. 4. As shown in Fig. 4, the training menu is set according to the following procedure: (1) Select the trainee's number (select the ID and name of the trainee from the pull-down menu). (2) Choose the shape of the mouth (e.g., "a" or "u") the trainee wants to train. (3) Choose the level of training (the degree of difficulty increases in ascending order from Level 1 to Level 3). (4) Select the number of times the trainee has received the training (automatic selection is possible). (5) Click the "Enter" button and go to the training page.
The training page is shown in Fig. 5. The expiration pattern of the therapist is indicated in the upper part and that of the trainee in the lower part of the figure.
The strength of expiration is discretely expressed as zero, weak, medium, or strong. The model expiration is indicated as a simple pattern because when changes in the volume of expiratory flow are shown in a graph, such as that in Fig. 3, it is difficult for trainees to imitate the pattern. Therefore, a simple graph was adopted in this study.
In Fig. 5, the thresholds of a mouth shaped "u" were set as 100, 400, and 700 (ml/s). When values measured by using the airflow sensor exceed the above thresholds, the level of expiratory flow is judged as weak, medium, or strong, respectively. These thresholds are adjusted on the basis of the trainees' condition, so that they can easily achieve the set goal, and motivation for training would increase.
Thresholds were determined by measuring the maximum expiratory flow X max of the trainee before starting training. Strong was then defined as more than 70% of X max , medium as more than 40% of X max , and weak as 10% of X max or more.
The therapist's expiration pattern is decided in advance, such that its length is 10 s. During the training, the indicator (vertical line) moves from left to right with the course of time. Trainees control the timing, strength, and length of their breath by checking the graph and the indicator.  As shown in Fig. 3, the performance characteristics of the sensor are such that there is an approximately 0.1 s delay at the start and finish of expiration. Therefore, assessment is not conducted during periods of sudden changes, such as at the start or end of expiration. The strength of the breath is assessed after sudden changes have ended.

Expiration training menus
Three levels of a therapist's expiration patterns are available, such that the trainees can choose an appropriate level.
As training advances, a therapist's expiration pattern becomes more refined with increases in the complexity of the interplay of changes in breath intensity and duration. A trainee proceeds to the next more complex level upon attaining advanced scores in the previous lower level of training.
The volume of expiratory flow differs on the basis of pronunciation using different mouth shapes. In this study, mouthpieces for the mouth shapes of "a" and "u" were prepared. Trainees choose either one of these and put on the mouthpiece corresponding to the selected pronunciation.

Evaluation of expiration training
Evaluation was conducted by calculating the score based on the differences in expiration patterns between the therapist and the trainee. Lower scores indicated better results.
The score was calculated using eq. Other formulas used were as follows. First, the rate of concordance (%) between the therapist's data and trainee's data was calculated using eq. (2).
Next, the gap of timing expiration (difference in mean time of starting position in ms) was calculated using eq. Furthermore, at each training session, not only the scores calculated with the formulas, but also the raw data of the expiratory flow were recorded, such that a future reevaluation would be possible. Recording with a time series facilitated confirming the effect of training.

Experimental Methods
Changes in training scores of healthy male adults (N = 5, A-E) using Level 1 therapist's data shown in Fig. 5 are displayed in Fig. 6. Each trainee conducted 10 training sessions.
Equations (2)-(5) were employed to derive scores on the vertical axis of Fig. 6. These scores represent differences between values presented by the therapist and values attained by the trainee with smaller scores representing improved results on the part of the trainee. As indicated in Fig. 6, improved scores serve to confirm progress through the step-wise training.
The feedback provided to the trainee is thought to better inform their adjustments in breath control (timing, strength, and length of expiration). In addition, successful reproduction of utterances requires repetitive training and the gamelike feedback in the form of a score may improve the level of engagement of the trainee.

Conclusions
The system developed in this study is a prototype and was tested with healthy adults. The results indicated an improvement of the score as the learning steps proceeded. Therefore, this system was considered effective for expiration training. However, longterm observations are necessary for confirming the improvement of expiration ability. It is suggested that in the future, this system should be tested and evaluated with people that actually require expiration training. Furthermore, adjustment of the airflow sensor (i.e., setting the thresholds for the strength of expiration) and the expiration pattern of therapists should be examined, so that trainees are able to more easily undergo the training. Additionally, this system could be used as a method to confirm the rehabilitative effects of training for strengthening respiration muscles, such as "stretching to expand the mobility range of the lungs" and "training for strengthening muscles around the lungs", among others. Furthermore, the addition of test items assessing basic capacities, such as lung capacity and volume of expiratory flow, to training items suggested in this study (e.g., timing and control of the length and strength of blowing), would also facilitate the confirmation of the effects of rehabilitation. Recently, the aging population has become a serious problem in Japan, and maintaining the health of the elderly has become important. It is suggested that in addition to providing speech training for people with disabilities, this system could be used to address new needs, such as speech rehabilitation and anti-aging training for the elderly. of Baby Science, the Japanese Society of Child Science, and the Japanese Society of Developmental Neuroscience. He is also the Editor-in-Chief of Baby Science and the Councilor of The Japanese Society of Child Neurology. He is working on several international collaborative studies of Developmental Coordination Disorder (DCD) with Canada, the Netherlands, Israel, UK, and Vietnam, as a representative of Japan in the International Society for Research into DCD.

About the Authors
Kiyoharu Tsuji is a physical therapist of the Fukui Prefectural Rehabilitation Center for Children with Disabilities. He is affiliated with the Japanese Physical Therapy Association, Rehabilitation Engineering Society of Japan, Japanese Society on Severe Motor and Intellectual Disabilities, and Japan Sport Association for the Disabled. He specializes in neurological physical therapy, life support and environmental adjustment physical therapy, and sport training for the disabled.
Yoshinori Mitsuhashi is a professor emeritus and currently a specially assigned professor of the Faculty of Education and Regional Studies at Fukui University, Japan. He graduated from Kwansei-Gakuin University, Department of Experimental Psychology in 1972, and finished his graduate study in 1980. He became an associate professor of Fukui University in 1981 and left as a professor in 2014. His research covers psychophysiology of cognition and education of handicapped children. In particular, he studies the cognitive process in the children and adults with neurodevelopmental disorders using the electrical activity of the brain.