The individual sounds used to create speech are called phonemes. Sources differ on exactly how many phonemes exist and what they are; some texts include sounds such as diphthongs (a combination of two vowel sounds) and affricatives (combinations of two consonants) as independent phonemes. In our work we model the 48 distinct phonemes identified in the ARPABET. [1]
Phonemes may be placed in one of three broad categories according to the mechanisms at work in their generation:
voiced sounds (vowels and pseudo-vowels)
fricative consonants
stop consonants