Time domain detection of pure vowel sounds for simplified computer interaction

This paper proposes the assessment of different methods designed for time domain detection of pure vowel sounds in order to develop future devices with simplified computer interaction focused on people with physical disabilities. The time domain detection method is based on filter the sound, detect periodic information and then compute the average periodic amplitude of the filtered and non-filtered signals. This detection method has been applied to pure Spanish vowels which have a similar pronunciation in different languages. The periodic information of the different vowels has been stored in reference databases in order to assess the different time domain detection methods proposed. The best detection results obtained showed a 100% detection success for the /a/, /e/, /i/ and /o/ phonemes and a 90% detection success for the /u/ phoneme when using the periodic time domain description of the signals.