← all emotions
Relief
Best adapter: rank 32 , checkpoint seen 5000 · EmoNet-z +0.73 (baseline 0.071) · blend 0.514 · genu 0.197 · speaker-sim 0.177.
Instruction prompt: GENERAL: A voice that is adult; extremely carrying alleviation; genuine in delivery.
SCRIPT:
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 -0.73 ±0.01 · bl 0.2 · gn 0.2 ▲ high -0.72
▼ low -0.75
z mean -0.49 ±0.49 · bl 0.4 · gn 0.1 ▲ high +0.49
▼ low -0.74
z mean +0.63 ±1.42 · bl 0.7 · gn 0.2 ▲ high +2.93
▼ low -0.64
with emo reference
z mean -0.74 ±0.00 · bl 0.2 · gn 0.3 ▲ high -0.73
▼ low -0.75
z mean -0.67 ±0.12 · bl 0.1 · gn 0.1 ▲ high -0.43
▼ low -0.74
z mean -0.45 ±0.34 · bl 0.2 · gn 0.3 ▲ high +0.01
▼ low -0.73
Sentence 2
no reference
z mean -0.42 ±0.27 · bl 0.4 · gn 0.3 ▲ high -0.14
▼ low -0.74
z mean -0.74 ±0.01 · bl 0.5 · gn 0.1 ▲ high -0.73
▼ low -0.74
z mean +0.01 ±0.96 · bl 0.5 · gn 0.2 ▲ high +1.62
▼ low -0.74
with emo reference
z mean -0.74 ±0.01 · bl 0.4 · gn 0.3 ▲ high -0.73
▼ low -0.76
z mean -0.63 ±0.12 · bl 0.3 · gn 0.1 ▲ high -0.46
▼ low -0.73
z mean -0.73 ±0.01 · bl 0.1 · gn 0.2 ▲ high -0.72
▼ low -0.74
Sentence 3
no reference
z mean -0.61 ±0.26 · bl 0.1 · gn 0.1 ▲ high -0.10
▼ low -0.77
z mean +1.37 ±1.27 · bl 0.4 · gn 0.1 ▲ high +2.91
▼ low -0.74
z mean +0.76 ±1.11 · bl 0.7 · gn 0.2 ▲ high +2.57
▼ low -0.64
with emo reference
z mean -0.74 ±0.01 · bl 0.1 · gn 0.1 ▲ high -0.73
▼ low -0.76
z mean -0.02 ±0.94 · bl 0.2 · gn 0.1 ▲ high +1.76
▼ low -0.74
z mean -0.74 ±0.02 · bl 0.2 · gn 0.3 ▲ high -0.71
▼ low -0.76
Sentence 4
no reference
z mean +0.24 ±0.49 · bl 0.1 · gn 0.1 ▲ high +0.78
▼ low -0.48
z mean -0.20 ±0.79 · bl 0.5 · gn 0.1 ▲ high +1.35
▼ low -0.70
z mean -0.36 ±0.72 · bl 0.5 · gn 0.2 ▲ high +1.08
▼ low -0.73
with emo reference
z mean +0.04 ±0.42 · bl 0.1 · gn 0.1 ▲ high +0.51
▼ low -0.73
z mean -0.39 ±0.68 · bl 0.2 · gn 0.1 ▲ high +0.97
▼ low -0.74
z mean -0.35 ±0.49 · bl 0.1 · gn 0.3 ▲ high +0.41
▼ low -0.75
Sentence 5
no reference
z mean -0.73 ±0.01 · bl 0.2 · gn 0.1 ▲ high -0.72
▼ low -0.74
z mean -0.55 ±0.27 · bl 0.5 · gn 0.1 ▲ high -0.03
▼ low -0.74
z mean +0.16 ±1.26 · bl 0.6 · gn 0.2 ▲ high +2.57
▼ low -0.70
with emo reference
z mean -0.73 ±0.00 · bl 0.1 · gn 0.1 ▲ high -0.73
▼ low -0.74
z mean -0.72 ±0.02 · bl 0.1 · gn 0.1 ▲ high -0.69
▼ low -0.75
z mean -0.74 ±0.01 · bl 0.1 · gn 0.2 ▲ high -0.73
▼ low -0.75