Consonantal Cues for Automatic Speech Recognition
Wilbur D. Larkin · The Journal of the Acoustical Society of America · 1960
A report is made on the investigation of 18 consonants (voiced and voiceless stops, affiricates, nasals /m/ and /n/, and fricatives /f, v, ð, θ, s, z, ȝ and ∫/) in combination with nine vowel phonemes /i, ɪ, ɛ, æ, a, ɔ, ᴜ, ʌ, and u/. This study is based upon both the transitions of the first four formants and the spectral distribution of energy of the stop and fricative bursts. The data for the investigation were obtained from subjects recording both consonant-vowel-consonant nonsense syllables and test words in sentences. The sentences were so constructed as to be identical except for the opposition of the consonant being investigated. The applicability of these data to consonants in middle and final position also is discussed. The results of this study and their application to an automatic speech recognition device are presented.