← all dimensions · markdown (agent skill) · manual home
Vulnerability VoiceNet dimension guide
Social & Interpersonalcode vn_VULNslot 50
This dimension measures the perceived openness or permeability 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.
Vulnerability controls the perceived openness or permeability of the speaker. The __high LoRA shifts it +0.05 at 125% (worst cost: burst-blend -0.26); the __low LoRA shifts it -0.40 at 125% (worst cost: burst-blend -0.36). The __low direction is the stronger, more reliable knob here.
▲ HIGH — Vulnerability __high
Dose-response
| dose | target | Δtarget | Δgenu | Δblend | Δqual |
|---|
| baseline (0%) | 0.523 | — | — | — | — |
| 25% | 0.573 | +0.05 | +0.04 | +0.03 | +0.04 |
| 50% | 0.512 | -0.01 | -0.01 | -0.03 | +0.04 |
| 75% | 0.556 | +0.03 | +0.01 | -0.08 | +0.00 |
| 100% | 0.543 | +0.02 | +0.06 | +0.03 | -0.11 |
| 125% ★ | 0.575 | +0.05 | -0.01 | -0.26 | -0.07 |
EmoNet emotion correlation
Pain+0.54
Helplessness+0.42
Sadness+0.36
Triumph-0.42
Amusement-0.41
Impatience and Irritability-0.40
VoiceNet dimension correlation
Disfluency+0.45
Respiration+0.44
Conversational Style+0.38
Arousal Shift-0.56
Harmonicity-0.55
Stance-0.49
Listen — 3 samples per dose
25%measured target: 0.51 · 0.58 · 0.50
50%measured target: 0.46 · 0.68 · 0.53
75%measured target: 0.35 · 0.51 · 0.52
100%measured target: 0.15 · 0.45 · 0.68
125% ★measured target: 0.21 · 0.44 · 0.55
Merging Vulnerability __high raises the dimension: at the recommended 125% dose the measured Vulnerability score moves from a 0.52 baseline to 0.57 (a +0.05 shift). The dose-response is mostly ordered but noisy between doses. That reach is a real cost — the biggest hit is to burst-blend (-0.26); across the board genuineness -0.01, burst-blend -0.26, speech quality -0.07. Emotionally it reads as Pain (r 0.54), Helplessness (0.42) and Sadness (0.36), and it moves away from Triumph, Amusement and Impatience and Irritability. On other voice dimensions it co-moves most with Disfluency (0.45), Respiration (0.44) and Conversational Style (0.38), while pulling down Arousal Shift, Harmonicity and Stance. Practical call: merge __high at 100–125%; expect Pain and Disfluency to rise and burst-blend to fall. If you don’t want Pain bleeding in, offset it with a prompt cue toward Triumph to keep the vulnerability shift without the side-effect.
▼ LOW — Vulnerability __low
Dose-response
| dose | target | Δtarget | Δgenu | Δblend | Δqual |
|---|
| baseline (0%) | 0.523 | — | — | — | — |
| 25% | 0.465 | -0.06 | -0.01 | -0.04 | +0.05 |
| 50% | 0.438 | -0.08 | -0.04 | -0.25 | -0.02 |
| 75% | 0.136 | -0.39 | -0.09 | -0.43 | -0.04 |
| 100% | 0.164 | -0.36 | -0.07 | -0.41 | -0.14 |
| 125% ★ | 0.128 | -0.40 | -0.05 | -0.36 | -0.24 |
EmoNet emotion correlation
Contemplation+0.64
Relief+0.64
Contentment+0.55
Triumph-0.72
Malevolence/Malice-0.57
Pride-0.57
VoiceNet dimension correlation
Cognitive Load+0.67
ASMR Style+0.66
Whisper-Talk Style+0.62
Harmonicity-0.85
Authoritative Style-0.85
Stance-0.85
Listen — 3 samples per dose
25%measured target: 0.18 · 0.64 · 0.38
50%measured target: 0.20 · 0.84 · 0.29
75%measured target: 0.09 · 0.00 · 0.01
100%measured target: 0.22 · 0.00 · 0.14
125% ★measured target: 0.28 · 0.14 · 0.00
Merging Vulnerability __low damps the dimension: at the recommended 125% dose the measured Vulnerability score moves from a 0.52 baseline to 0.13 (a -0.40 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 (-0.36); across the board genuineness -0.05, burst-blend -0.36, speech quality -0.24. Emotionally it reads as Contemplation (r 0.64), Relief (0.64) and Contentment (0.55), and it moves away from Triumph, Malevolence/Malice and Pride. On other voice dimensions it co-moves most with Cognitive Load (0.67), ASMR Style (0.66) and Whisper-Talk Style (0.62), while pulling down Harmonicity, Authoritative Style and Stance. Practical call: merge __low at 100–125%; expect Contemplation and Cognitive Load to rise and burst-blend to fall. If you don’t want Contemplation bleeding in, offset it with a prompt cue toward Triumph to keep the vulnerability shift without the side-effect.