An emotional detection technique which extracts acoustic features from audio recordings of speech was developed. Though the formant frequency of individual voices may contribute to emotional variations in speech, the differences between vowels has an influence on feature extraction. To reduce the influence, a simple procedure was developed to extract relative features of vowels for every mora. The estimation performance of this emotional detection technique was improved by 11\% using relative formant frequencies instead of formant frequencies. The strengths of some emotional expressions were also reflected in some features. The effectiveness of using acoustic features to estimate the category of emotionally inflected speech was confirmed.
JavaScript jest wyłączony w Twojej przeglądarce internetowej. Włącz go, a następnie odśwież stronę, aby móc w pełni z niej korzystać.