TTS 한국어 합성음의 속도와 피치가 정보 인지 및 감성에 미치는 영향 평가: 시각장애인을 대상으로

Abstract

The purpose of this study was to find the maximum listening speed and optimal pitch level of Korean TTS(Text-to-speech) for the blind. TTS is called a screen reader and synthesis program that reads the text to speech. TTS is one of the representative AT(Assitive Technology) for the visually impairment. The blind people use the device depending on the TTS voice. Recently TTS is mounted on various smart devices as well as PC. In the case of smart devices, the blind people should rely more on the voice, because most smart devices are touch interface. The sound can affect the evaluation of the products as well as the user's affect. The blind people have a desire to auditory stimuli, and they may more sensitive. However it is difficult to find relevant research. The process of obtaining the information over the TTS can be influenced by the rate, pitch, volume, stress, pause of voice. The prosodic elements containing the stress and pause have been considered in oder to make the natural voice. However studies on the rate and pitch are difficult to find. The pitch is different from the intonation. The rate and pitch can affect to transfer information. In particular, the blind people have affective needs for the pitch. In most TTS program, users can control level of rate and pitch. But the range of speed control is difficult to determine as to whether adequate. This study conducted the experiments in order to find the maximum listening speed and evaluate effect of the pitch on user's affect. The experiments were designed to consider the actual use. 24 blind participants took part in this study. They are total blindness and have used TTS more than 5 years. The experiments are divided into two evaluation. One is about finding listening speed and the other is affect evaluation. To find maximum listening speed, users listen words and sentences. After listening two words with two syllables, users determine that the same words or different words. Also users listen a sentence. This study evaluate whether users listen a key word in a sentence. In affect evaluation, users score the voice using 100-point scale. The evaluation contains 5 affect elements(loudness, liveness, comfortable, attractiveness, overall satisfaction). In most TTS program, the maximum speed is about 850 SPM. However the control range is not sufficient. In this study, users can listen the information at a faster rate. In the case of affect evaluation, this study contain 3 levels of pitch considering Korean female voice. As a result, the pitch of women in their 20s has a positive influence on affect. the pitch is about 220 Hz.