Difference between revisions of "Speech Recognition for Education"
Jump to navigation
Jump to search
| Line 8: | Line 8: | ||
*a model (the SpeechRater) automatically scoring open-ended spoken responses for speakers with documented or suspected speech impairments | *a model (the SpeechRater) automatically scoring open-ended spoken responses for speakers with documented or suspected speech impairments | ||
*SpeechRater was less accurate for test takers who were deferred for signs of speech impairment (ρ<sup>2</sup> = .57) than test takers who were given accommodations for documented disabilities (ρ<sup>2</sup> = .73) | *SpeechRater was less accurate for test takers who were deferred for signs of speech impairment (ρ<sup>2</sup> = .57) than test takers who were given accommodations for documented disabilities (ρ<sup>2</sup> = .73) | ||
Loukina et al. (2019) [[https://aclanthology.org/W19-4401.pdf pdf]] | |||
*Models providing automated speech scores on English language proficiency assessment | |||
*L1-specific model trained on the speaker’s native language was the least fair, especially for Chinese, Japanese, and Korean speakers, but not for German speakers | |||
*All models (Baseline, Fair feature subset, L1-specific) performed disadvantageously for Japanese speakers | |||
Revision as of 16:31, 28 March 2022
Wang et al. (2018) [pdf]
- Automated scoring model for evaluating English spoken responses
- SpeechRater gave a significantly lower score than human raters for German
- SpeechRater scored in favor of Chinese group, with H1-rater scores higher than mean
Loukina & Buzick (2017) [pdf]
- a model (the SpeechRater) automatically scoring open-ended spoken responses for speakers with documented or suspected speech impairments
- SpeechRater was less accurate for test takers who were deferred for signs of speech impairment (ρ2 = .57) than test takers who were given accommodations for documented disabilities (ρ2 = .73)
Loukina et al. (2019) [pdf]
- Models providing automated speech scores on English language proficiency assessment
- L1-specific model trained on the speaker’s native language was the least fair, especially for Chinese, Japanese, and Korean speakers, but not for German speakers
- All models (Baseline, Fair feature subset, L1-specific) performed disadvantageously for Japanese speakers