Table 2. Correlations between human- and computer-generated ratings within conditions.
Condition | Correlation [95% CI] | Number of recordings | ||||
---|---|---|---|---|---|---|
r | ICC(1) | Training | Test | |||
(+) | (–) | (+) | (–) | |||
Enhance | .81 [.78, .84] | .64 [.59, .68] | .79 [.76, .82] | .61 [.55, .66] | 1,047 | 569 |
Normal | .81 [.78, .84] | .55 [.49, .61] | .79 [.76, .82] | .49 [.42, .55] | 880 | 516 |
Suppress | .85 [.83, .87] | .44 [.38, .51] | .83 [.80, .85] | .35 [.28, .42] | 1,040 | 596 |
Notes. (+) = positive valence ratings; (–) = negative valence ratings; r = Pearson’s correlation; ICC = Intraclass correlation coefficient. All results reported are on test sets.