ISCA Workshop on Plasticity in Speech Perception (PSP2005)
Senate House, London, UK
Preliminary results and a proposed Speech Perception Assessment and Training System (SPATS) for the aural rehabilitation of adult hearingaid and cochlear-implant users are described. The proposed system targets bottom-up, or analytic, speech-perception abilities, and has two parts: an assessment component and a training component. The assessment component provides an inventory of a client's abilities to identify the constituents of English syllables, that is, syllable onsets, nuclei, and codas in quiet and in background noise. The training algorithm adaptively selects items and noise levels so that practice is concentrated on items of moderate difficulty weighted by their importance for speech understanding. Item importance is estimated from frequency of occurrence data. Further, learning is enhanced by providing many practice trials, immediate feedback, and, following errors, rapid comparisons of intended sounds with their confusors. To encourage transfer to real-world speech, brief periods of training on words and phrases are interspersed with the training on syllable constituents. Validation studies will measure how well the assessment and training system predicts and improves general success in speech communication. In a preliminary study, five experienced hearing-aid users with sensorineural hearing loss were given 14 hours of intensive training identifying syllable-initial consonants in quiet and noise. Their performance was compared to that of five similar hearing-aid users with no special training. All listeners had moderate to severe hearing losses and had worn hearing aids for at least one year. All were pre-tested with a set of 20 consonants combined with three vowels /i,a,u/ as spoken by six different talkers. Pretests were conducted in quiet and in noise (multi-talker babble) at moderate signal-to-noise ratios (SNR's). Training was conducted with eight target consonants (TCs). The TCs were in each listener's middle range of difficulty and the three most common confusors for each target were individually selected forming target sets of four consonants. Training was conducted in quiet and noise. During training, trial-by-trial feedback was given and, following an error, the listener could rapidly compare the intended syllable with its confusor. In noise, the SNR adapted to a criterion of 80% correct. There were no differences between Training and Control listeners on the Pretests. After training, there was a significant fivepercent advantage for the trained listeners. Training generalized to talkers never heard during training. Graphs depicting the progress of adaptive training in noise are presented. [Supported by NIDCD]
Bibliographic reference. Miller, James / Watson, Charles / Dalby, Jonathan / Burleson, Deborah (2005): "Training experienced hearing-aid users to identify syllable constituents in quiet and noise", In PSP2005, 237 (abstract).