Analysis of emotional expression by visualization of the human and synthesized speech signal sets — A consideration of audio-visual advantage-

In recent years, speech synthesis technology has been developed, but the method of expressing emotions with synthesized speech was one of the issue. Therefore, in the latest speech synthesis technology, a method for expressing emotion by synthesized speech continues to be developed. For example, as one of them, it is a method of changing the physical values such as the height of the sound and the speaking speed, reproducing emotions such as joy and anger, and reading out the text. Emotional expression of synthesized speech is to reproduce human emotions by physically changing the height of sound, speech speed, intonation, and the like. However, we do not hear much comparison between synthesized speech expressing emotion and human one. Therefore, in this study, we conducted a control experiment on whether there is a difference between synthesized speech expressing emotion and human speech. In this control experiment method, two sounds, one synthesized by using one of the speech synthesis software and one recorded with human speech, are visualized as speech signals, and the differences that can be analyzed from these characteristics are listed. In this research, the speech signal is displayed as a waveform, frequency analysis and so on. As a result, various speech features were able to be grasped. For example, in the voice of anger by humans, as the anger becomes stronger, the strength of the vowel becomes stronger and the overall volume increases, it turned out that the synthesized speech also reproduced by imitating the feature.