← all dimensions · markdown (agent skill) · manual home

Arousal VoiceNet dimension guide

Emotion & Affectcode vn_AROUslot 55
This dimension measures the physiological “engine speed” of the speaker. Below: the merge-scaling dose-response for the __high and __low LoRA (laion/moss-voicenet-dimension-loras) at 25/50/75/100/125%, the measured target value at each dose, the top-3 ± correlated EmoNet emotions and VoiceNet dimensions, three audio samples per dose, and a grounded prompting tip. Costs are shifts vs. the 24-sample neutral baseline in genuineness, vocal-burst blend and speech quality.

Arousal controls the physiological “engine speed” of the speaker. The __high LoRA shifts it +0.73 at 100% (worst cost: burst-blend -0.43); the __low LoRA shifts it -0.09 at 50% (worst cost: burst-blend -0.23). The __high direction is the stronger, more reliable knob here.
▲ HIGH — Arousal __high
Dose-response
dosetargetΔtargetΔgenuΔblendΔqual
baseline (0%)0.143————
25%0.235+0.09-0.04-0.17-0.06
50%0.307+0.16+0.00-0.16-0.04
75%0.540+0.40-0.03-0.33-0.10
100% ★0.874+0.73-0.08-0.43-0.23
125%0.796+0.65-0.09-0.44-0.11
EmoNet emotion correlation
Triumph+0.72
Anger+0.68
Bitterness+0.61
Contemplation-0.70
Relief-0.65
Contentment-0.64
VoiceNet dimension correlation
Attack+0.98
Metallic Character+0.98
Stance+0.94
Vulnerability-0.86
Whisper-Talk Style-0.79
Warmth-0.79
Listen — 3 samples per dose
25%measured target: 0.32 · 0.31 · 0.17
50%measured target: 0.20 · 0.67 · 0.22
75%measured target: 0.14 · 0.62 · 0.65
100% ★measured target: 0.82 · 0.94 · 0.91
125%measured target: 0.85 · 0.78 · 0.83
Merging Arousal __high ramps the dimension: at the recommended 100% dose the measured Arousal score moves from a 0.14 baseline to 0.87 (a +0.73 shift). The dose-response is clean and monotonic — each step pushes the dimension further. That reach is a steep cost — the biggest hit is to burst-blend (-0.43); across the board genuineness -0.08, burst-blend -0.43, speech quality -0.23. Emotionally it reads as Triumph (r 0.72), Anger (0.68) and Bitterness (0.61), and it moves away from Contemplation, Relief and Contentment. On other voice dimensions it co-moves most with Attack (0.98), Metallic Character (0.98) and Stance (0.94), while pulling down Vulnerability, Whisper-Talk Style and Warmth. Practical call: merge __high at 75–100%; expect Triumph and Attack to rise and burst-blend to fall. If you don’t want Triumph bleeding in, offset it with a prompt cue toward Contemplation to keep the arousal shift without the side-effect.
▼ LOW — Arousal __low
Dose-response
dosetargetΔtargetΔgenuΔblendΔqual
baseline (0%)0.143————
25%0.076-0.07+0.02+0.07-0.03
50% ★0.055-0.09-0.01-0.23-0.07
75%0.061-0.08-0.07-0.22-0.01
100%0.141+0.00-0.04-0.31-0.06
125%0.118-0.03-0.07-0.25+0.05
EmoNet emotion correlation
Jealousy & Envy+0.58
Teasing+0.54
Bitterness+0.35
Sourness-0.36
Fatigue/Exhaustion-0.30
Contentment-0.24
VoiceNet dimension correlation
Metallic Character+0.82
Attack+0.73
Stance+0.72
Conversational Style-0.46
Vulnerability-0.34
ASMR Style-0.32
Listen — 3 samples per dose
25%measured target: 0.00 · 0.16 · 0.13
50% ★measured target: 0.04 · 0.08 · 0.01
75%measured target: 0.11 · 0.08 · 0.07
100%measured target: 0.12 · 0.02 · 0.15
125%measured target: 0.15 · 0.15 · 0.11
Merging Arousal __low damps the dimension: at the recommended 50% dose the measured Arousal score moves from a 0.14 baseline to 0.06 (a -0.09 shift). The dose-response is mostly ordered but noisy between doses. That reach is a real cost — the biggest hit is to burst-blend (-0.23); across the board genuineness -0.01, burst-blend -0.23, speech quality -0.07. Emotionally it reads as Jealousy & Envy (r 0.58), Teasing (0.54) and Bitterness (0.35), and it moves away from Sourness, Fatigue/Exhaustion and Contentment. On other voice dimensions it co-moves most with Metallic Character (0.82), Attack (0.73) and Stance (0.72), while pulling down Conversational Style, Vulnerability and ASMR Style. Practical call: merge __low at 25–75%; expect Jealousy & Envy and Metallic Character to rise and burst-blend to fall. If you don’t want Jealousy & Envy bleeding in, offset it with a prompt cue toward Sourness to keep the arousal shift without the side-effect.
Data: MOSS-VA-v2 base + per-dimension LoRA, neutral text, no reference · 5 merge doses × 8 scored samples/dose · correlations are Pearson r over the sample pool · scaling grid · LoRA repo