← all emotions
Sadness
Best adapter: rank 32 , checkpoint seen 3336 · EmoNet-z +1.90 (baseline 0.731) · blend 0.593 · genu 0.262 · speaker-sim 0.109.
Instruction prompt: GENERAL: Melancholy and soft, tears just behind every word.
SCRIPT:
(on the verge of tears) The train leaves at seven. (barely holding back tears) If we don't leave now, we'll miss it.
Five neutral sentences, each in English / Deutsch / Français, no reference vs with emotion reference (a same-emotion training clip supplies the voice; the prompt applies the emotion).
Each cell = 5 seeds. We show mean±std of the emotion score and play the highest and lowest seed (top = strongest emotion, bottom = weakest), since a fixed seed pins the sampling and single clips vary.
Sentence 1
no reference
z mean -0.55 ±0.55 · bl 0.4 · gn 0.2 ▲ high +0.53
▼ low -0.87
z mean -0.87 ±0.00 · bl 0.5 · gn 0.1 ▲ high -0.87
▼ low -0.87
z mean +0.62 ±0.52 · bl 0.8 · gn 0.2 ▲ high +1.41
▼ low +0.02
with emo reference
z mean -0.33 ±0.73 · bl 0.3 · gn 0.5 ▲ high +0.97
▼ low -0.87
z mean -0.75 ±0.24 · bl 0.1 · gn 0.5 ▲ high -0.28
▼ low -0.87
z mean +1.10 ±0.78 · bl 0.3 · gn 0.5 ▲ high +2.17
▼ low -0.08
Sentence 2
no reference
z mean -0.87 ±0.00 · bl 0.8 · gn 0.3 ▲ high -0.87
▼ low -0.87
z mean -0.87 ±0.00 · bl 0.5 · gn 0.1 ▲ high -0.87
▼ low -0.87
z mean +0.18 ±1.02 · bl 0.6 · gn 0.3 ▲ high +1.70
▼ low -0.87
with emo reference
z mean -0.25 ±0.61 · bl 0.3 · gn 0.6 ▲ high +0.78
▼ low -0.87
z mean +0.82 ±0.77 · bl 0.1 · gn 0.5 ▲ high +2.27
▼ low +0.21
z mean +0.24 ±1.14 · bl 0.3 · gn 0.6 ▲ high +2.23
▼ low -0.87
Sentence 3
no reference
z mean -0.31 ±0.66 · bl 0.2 · gn 0.1 ▲ high +0.92
▼ low -0.87
z mean +0.06 ±0.67 · bl 0.5 · gn 0.2 ▲ high +0.98
▼ low -0.87
z mean +0.85 ±0.63 · bl 0.6 · gn 0.2 ▲ high +1.57
▼ low -0.01
with emo reference
z mean -0.07 ±0.55 · bl 0.3 · gn 0.5 ▲ high +0.58
▼ low -0.87
z mean +0.99 ±0.86 · bl 0.1 · gn 0.4 ▲ high +2.55
▼ low +0.02
z mean +0.57 ±1.03 · bl 0.1 · gn 0.5 ▲ high +2.13
▼ low -0.87
Sentence 4
no reference
z mean -0.67 ±0.25 · bl 0.4 · gn 0.1 ▲ high -0.27
▼ low -0.87
z mean -0.53 ±0.67 · bl 0.5 · gn 0.1 ▲ high +0.80
▼ low -0.87
z mean -0.87 ±0.00 · bl 0.7 · gn 0.3 ▲ high -0.87
▼ low -0.87
with emo reference
z mean +0.20 ±0.67 · bl 0.2 · gn 0.6 ▲ high +1.10
▼ low -0.87
z mean +0.60 ±1.35 · bl 0.3 · gn 0.5 ▲ high +2.69
▼ low -0.87
z mean +0.91 ±0.52 · bl 0.3 · gn 0.5 ▲ high +1.70
▼ low +0.21
Sentence 5
no reference
z mean -0.31 ±0.66 · bl 0.3 · gn 0.2 ▲ high +0.66
▼ low -0.87
z mean -0.85 ±0.04 · bl 0.5 · gn 0.2 ▲ high -0.77
▼ low -0.87
z mean +0.10 ±1.19 · bl 0.6 · gn 0.2 ▲ high +1.56
▼ low -0.87
with emo reference
z mean -0.63 ±0.45 · bl 0.2 · gn 0.6 ▲ high +0.27
▼ low -0.87
z mean +0.30 ±0.81 · bl 0.1 · gn 0.4 ▲ high +1.32
▼ low -0.87
z mean +0.65 ±1.29 · bl 0.3 · gn 0.5 ▲ high +2.32
▼ low -0.87