Perception of synthetic speech sounds by hearing-impaired listeners

Volunteers (7 men, 8 women) with mild to severe hearing losses listened by monaural earphone to 5 series of computer-synthesized speech syllables, and identified each token. Presentation level 30 db SL. The 5 series consisted of (a, b) steady-state vowels varying from to [o] and from [o] to [a], (c,d) CV syllables varying from [ba] to [wa], and from [da] to [ya], differing in the rate of formant transitions at onset, and (e) CV syllables varying from [ba] to [da] differing in both direction and extent of formant transitions from locus to vowel. All hearing-impaired Ss categorized the vowels essentially the same as normal control Ss; some had difficulty identifying the stop vs glide series, and all but 2 were unable to identify the [b-d] series normally. A few impaired Ss improved at a presentation level close to tolerable limits. Results imply a "hierarchy of difficulty" based on the nature of the spectral cue in different speech sound types.