LISTENER PREFERENCE RATINGS FOR AAC SPEECH.

Judges rated voice features for nine speech samples representing synthesized and natural speech. Among features rated were quality, prosody, pitch, rate, and understandability. Synthesized speech samples were those typically used in communication devices, (e.g., DecTalk, Smoothtalker). Samples representing natural speech were those produced by a human speaker, (e.g., normal speech, alaryngeal speech). Subjects for the speech samples were recruited from licensed/certified speech language pathologists who were specialists in voice/alaryngeal speech. Results will be discussed in terms of ranked preferences and significant differences in ratings of the speech features, as well as implications for use in AAC. [ABSTRACT FROM AUTHOR]