← all dimensions · markdown (agent skill) · manual home
Volatility VoiceNet dimension guide
Emotion & Affectcode vn_VOLTslot 56
This dimension measures how much the emotional state shifts or vibrates within the clip. 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.
Volatility controls how much the emotional state shifts or vibrates within the clip. The __high LoRA shifts it +0.28 at 125% (worst cost: speech quality -0.37); the __low LoRA shifts it -0.15 at 75% (worst cost: burst-blend -0.12). The __high direction is the stronger, more reliable knob here.
▲ HIGH — Volatility __high
Dose-response
| dose | target | Δtarget | Δgenu | Δblend | Δqual |
|---|
| baseline (0%) | 0.398 | — | — | — | — |
| 25% | 0.420 | +0.02 | -0.04 | -0.12 | +0.03 |
| 50% | 0.583 | +0.19 | +0.00 | -0.14 | -0.11 |
| 75% | 0.607 | +0.21 | -0.03 | -0.33 | -0.22 |
| 100% | 0.602 | +0.20 | -0.04 | -0.28 | -0.21 |
| 125% ★ | 0.676 | +0.28 | +0.02 | -0.12 | -0.37 |
EmoNet emotion correlation
Triumph+0.59
Anger+0.57
Impatience and Irritability+0.53
Contemplation-0.64
Affection-0.58
Concentration-0.51
VoiceNet dimension correlation
Dramatic Style+0.82
Dynamic Arc+0.78
Arousal+0.77
Warmth-0.82
Monologue Style-0.78
Whisper-Talk Style-0.77
Listen — 3 samples per dose
25%measured target: 0.28 · 0.59 · 0.42
50%measured target: 0.46 · 0.30 · 0.56
75%measured target: 0.26 · 0.60 · 0.55
100%measured target: 0.59 · 0.64 · 0.66
125% ★measured target: 0.72 · 0.77 · 0.62
Merging Volatility __high drives the dimension: at the recommended 125% dose the measured Volatility score moves from a 0.40 baseline to 0.68 (a +0.28 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 speech quality (-0.37); across the board genuineness +0.02, burst-blend -0.12, speech quality -0.37. Emotionally it reads as Triumph (r 0.59), Anger (0.57) and Impatience and Irritability (0.53), and it moves away from Contemplation, Affection and Concentration. On other voice dimensions it co-moves most with Dramatic Style (0.82), Dynamic Arc (0.78) and Arousal (0.77), while pulling down Warmth, Monologue Style and Whisper-Talk Style. Practical call: merge __high at 100–125%; expect Triumph and Dramatic Style to rise and speech quality to fall. If you don’t want Triumph bleeding in, offset it with a prompt cue toward Contemplation to keep the volatility shift without the side-effect.
▼ LOW — Volatility __low
Dose-response
| dose | target | Δtarget | Δgenu | Δblend | Δqual |
|---|
| baseline (0%) | 0.398 | — | — | — | — |
| 25% | 0.328 | -0.07 | -0.04 | +0.03 | +0.07 |
| 50% | 0.330 | -0.07 | -0.08 | -0.09 | +0.12 |
| 75% ★ | 0.248 | -0.15 | -0.09 | -0.12 | +0.14 |
| 100% | 0.286 | -0.11 | -0.13 | -0.19 | +0.15 |
| 125% | 0.255 | -0.14 | -0.06 | -0.14 | +0.13 |
EmoNet emotion correlation
Jealousy & Envy+0.37
Doubt+0.36
Elation+0.36
Emotional Numbness-0.43
Contemplation-0.39
Embarrassment-0.29
VoiceNet dimension correlation
Casual Style+0.71
Respiration+0.61
Genuineness+0.55
Valence Shift-0.75
Structure-0.69
Teacher/Didactic Style-0.67
Listen — 3 samples per dose
25%measured target: 0.34 · 0.40 · 0.25
50%measured target: 0.20 · 0.34 · 0.26
75% ★measured target: 0.21 · 0.17 · 0.28
100%measured target: 0.17 · 0.19 · 0.34
125%measured target: 0.33 · 0.24 · 0.22
Merging Volatility __low suppresses the dimension: at the recommended 75% dose the measured Volatility score moves from a 0.40 baseline to 0.25 (a -0.15 shift). The dose-response is monotonic but it saturates early, so higher doses add little. That reach is a modest cost — the biggest hit is to burst-blend (-0.12); across the board genuineness -0.09, burst-blend -0.12, speech quality +0.14. Emotionally it reads as Jealousy & Envy (r 0.37), Doubt (0.36) and Elation (0.36), and it moves away from Emotional Numbness, Contemplation and Embarrassment. On other voice dimensions it co-moves most with Casual Style (0.71), Respiration (0.61) and Genuineness (0.55), while pulling down Valence Shift, Structure and Teacher/Didactic Style. Practical call: merge __low at 50–100%; expect Jealousy & Envy and Casual Style to rise and burst-blend to fall. If you don’t want Jealousy & Envy bleeding in, offset it with a prompt cue toward Emotional Numbness to keep the volatility shift without the side-effect.