Related Experiment Videos
Discrimination of computer-synthesized speech.
Perceptual and Motor Skills
|October 1, 1984
Summary
Learning improves computer-synthesized speech discrimination. Correcting errors significantly enhances understanding of synthesized monosyllables, though performance remains below human speech quality.
Area of Science:
- Speech processing
- Human-computer interaction
- Auditory learning
Background:
- Computer-synthesized speech (CSS) is increasingly prevalent.
- Understanding CSS is crucial for effective human-computer interaction.
- Limited research exists on the learning effects of CSS discrimination.
Purpose of the Study:
- To investigate the impact of error correction on learning to discriminate computer-synthesized monosyllabic words.
- To compare the learning effectiveness between groups receiving error correction and those not.
Main Methods:
- 100 computer-produced monosyllabic words were presented to 30 adult subjects in two groups.
- One group received immediate correction for their discrimination errors.
- The second group received no error correction.
Main Results:
- The group with corrected errors showed significant improvement in discriminating CSS over the second 50 presentations.
- The uncorrected group did not exhibit significant learning gains.
- Discrimination of CSS, even with correction, remained approximately 14% poorer than human speech.
Conclusions:
- Error correction is a key factor in improving the discrimination of computer-synthesized speech.
- Learners can adapt to CSS, but its intelligibility still lags behind natural human speech.
- Further research is needed to bridge the performance gap between CSS and human speech discrimination.