Robustness spectral tilt

Overall scores

	w2v2-L	hubert-L	wavlm	data2vec
Overall Score	87.5% passed tests (7 passed / 1 failed).	100.0% passed tests (8 passed / 0 failed).	75.0% passed tests (6 passed / 2 failed).	62.5% passed tests (5 passed / 3 failed).

Change Ccc Downward Tilt

Threshold: -0.05

Data	Change CCC Downward Tilt
	w2v2-L	hubert-L	wavlm	data2vec
iemocap-2.3.0-emotion.dimensions.test.gold_standard	-0.01	-0.01	-0.02	-0.01
msppodcast-2.6.1-emotion.dimensions.test-1.gold_standard	0.00	0.00	-0.01	0.01
mean	-0.01	-0.01	-0.01	0.00

Change Ccc Upward Tilt

Threshold: -0.05

Data	Change CCC Upward Tilt
	w2v2-L	hubert-L	wavlm	data2vec
iemocap-2.3.0-emotion.dimensions.test.gold_standard	-0.02	-0.02	0.02	-0.04
msppodcast-2.6.1-emotion.dimensions.test-1.gold_standard	-0.02	-0.02	-0.03	-0.07
mean	-0.02	-0.02	-0.00	-0.06

Percentage Unchanged Predictions Downward Tilt

Threshold: 0.8

Data	Percentage Unchanged Predictions Downward Tilt
	w2v2-L	hubert-L	wavlm	data2vec
iemocap-2.3.0-emotion.dimensions.test.gold_standard	0.99	1.00	1.00	0.97
msppodcast-2.6.1-emotion.dimensions.test-1.gold_standard	0.99	0.99	0.91	0.91
mean	0.99	0.99	0.96	0.94

Percentage Unchanged Predictions Upward Tilt

Threshold: 0.8

Data	Percentage Unchanged Predictions Upward Tilt
	w2v2-L	hubert-L	wavlm	data2vec
iemocap-2.3.0-emotion.dimensions.test.gold_standard	0.73	0.86	0.68	0.49
msppodcast-2.6.1-emotion.dimensions.test-1.gold_standard	0.88	0.91	0.79	0.68
mean	0.80	0.89	0.74	0.58

Visualization Downward Tilt

Difference of predictions for original audio and audio with a downward spectral tilt. The allowed prediction difference \(\delta < 0.05\) is highlighted in green in the upper plot. The lower plot shows the distributions of the two predictions.

w2v2-L	hubert-L	wavlm	data2vec