← all dimensions · markdown (agent skill) · manual home
Velocity Flux VoiceNet dimension guide
Temporal Dynamicscode vn_VFLXslot 68
This dimension measures the change in speaking speed over the duration of 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.
Velocity Flux controls the change in speaking speed over the duration of the clip. The __high LoRA shifts it +0.42 at 125% (worst cost: burst-blend -0.41); the __low LoRA shifts it -0.01 at 100% (worst cost: speech quality -0.14). The __high direction is the stronger, more reliable knob here.
▲ HIGH — Velocity Flux __high
Dose-response
| dose | target | Δtarget | Δgenu | Δblend | Δqual |
|---|
| baseline (0%) | 0.261 | — | — | — | — |
| 25% | 0.340 | +0.08 | -0.05 | +0.02 | +0.04 |
| 50% | 0.433 | +0.17 | -0.07 | -0.09 | +0.08 |
| 75% | 0.680 | +0.42 | -0.10 | -0.32 | -0.01 |
| 100% | 0.678 | +0.42 | -0.10 | -0.33 | -0.07 |
| 125% ★ | 0.684 | +0.42 | -0.09 | -0.41 | -0.17 |
EmoNet emotion correlation
Triumph+0.54
Amusement+0.50
Impatience and Irritability+0.49
Longing-0.55
Relief-0.51
Fatigue/Exhaustion-0.46
VoiceNet dimension correlation
Pitch Range+0.84
Head Resonance+0.82
Dynamic Arc+0.81
Cognitive Load-0.82
Warmth-0.67
Vulnerability-0.66
Listen — 3 samples per dose
25%measured target: 0.23 · 0.17 · 0.55
50%measured target: 0.35 · 0.48 · 0.31
75%measured target: 0.81 · 0.60 · 0.72
100%measured target: 0.37 · 0.64 · 0.69
125% ★measured target: 0.75 · 0.74 · 0.68
Merging Velocity Flux __high raises the dimension: at the recommended 125% dose the measured Velocity Flux score moves from a 0.26 baseline to 0.68 (a +0.42 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 (-0.41); across the board genuineness -0.09, burst-blend -0.41, speech quality -0.17. Emotionally it reads as Triumph (r 0.54), Amusement (0.50) and Impatience and Irritability (0.49), and it moves away from Longing, Relief and Fatigue/Exhaustion. On other voice dimensions it co-moves most with Pitch Range (0.84), Head Resonance (0.82) and Dynamic Arc (0.81), while pulling down Cognitive Load, Warmth and Vulnerability. Practical call: merge __high at 100–125%; expect Triumph and Pitch Range to rise and burst-blend to fall. If you don’t want Triumph bleeding in, offset it with a prompt cue toward Longing to keep the velocity flux shift without the side-effect.
▼ LOW — Velocity Flux __low
Dose-response
| dose | target | Δtarget | Δgenu | Δblend | Δqual |
|---|
| baseline (0%) | 0.261 | — | — | — | — |
| 25% | 0.418 | +0.16 | -0.03 | +0.02 | +0.03 |
| 50% | 0.290 | +0.03 | +0.02 | -0.10 | -0.01 |
| 75% | 0.343 | +0.08 | -0.02 | -0.20 | -0.13 |
| 100% ★ | 0.255 | -0.01 | +0.06 | -0.12 | -0.14 |
| 125% | 0.297 | +0.04 | +0.01 | -0.07 | -0.24 |
EmoNet emotion correlation
Concentration+0.38
Contemplation+0.31
Emotional Numbness+0.22
Fatigue/Exhaustion-0.43
Bitterness-0.32
Sadness-0.31
VoiceNet dimension correlation
Head Resonance+0.64
Teacher/Didactic Style+0.60
ASMR Style+0.58
Genuineness-0.58
Cognitive Load-0.51
Perceived Gender-0.44
Listen — 3 samples per dose
25%measured target: 0.53 · 0.49 · 0.23
50%measured target: 0.20 · 0.34 · 0.34
75%measured target: 0.49 · 0.39 · 0.19
100% ★measured target: 0.34 · 0.29 · 0.00
125%measured target: 0.45 · 0.35 · 0.26
Merging Velocity Flux __low suppresses the dimension: at the recommended 100% dose the measured Velocity Flux score moves from a 0.26 baseline to 0.26 (a -0.01 shift). The dose-response is mostly ordered but noisy between doses. That reach is a modest cost — the biggest hit is to speech quality (-0.14); across the board genuineness +0.06, burst-blend -0.12, speech quality -0.14. Emotionally it reads as Concentration (r 0.38), Contemplation (0.31) and Emotional Numbness (0.22), and it moves away from Fatigue/Exhaustion, Bitterness and Sadness. On other voice dimensions it co-moves most with Head Resonance (0.64), Teacher/Didactic Style (0.60) and ASMR Style (0.58), while pulling down Genuineness, Cognitive Load and Perceived Gender. Practical call: merge __low at 75–100%; expect Concentration and Head Resonance to rise and speech quality to fall. If you don’t want Concentration bleeding in, offset it with a prompt cue toward Fatigue/Exhaustion to keep the velocity flux shift without the side-effect.