← all dimensions · markdown (agent skill) · manual home
Dynamic Arc VoiceNet dimension guide
Temporal Dynamicscode vn_DARCslot 69
This dimension measures the trajectory of loudness across 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.
Dynamic Arc controls the trajectory of loudness across the clip. The __high LoRA shifts it +0.63 at 125% (worst cost: burst-blend -0.42); the __low LoRA shifts it -0.02 at 25% (worst cost: genuineness -0.01). The __high direction is the stronger, more reliable knob here.
▲ HIGH — Dynamic Arc __high
Dose-response
| dose | target | Δtarget | Δgenu | Δblend | Δqual |
|---|
| baseline (0%) | 0.022 | — | — | — | — |
| 25% | 0.112 | +0.09 | -0.03 | -0.13 | +0.02 |
| 50% | 0.219 | +0.20 | -0.04 | -0.29 | -0.04 |
| 75% | 0.560 | +0.54 | -0.09 | -0.37 | -0.09 |
| 100% | 0.643 | +0.62 | -0.09 | -0.41 | -0.16 |
| 125% ★ | 0.653 | +0.63 | -0.08 | -0.42 | -0.18 |
EmoNet emotion correlation
Triumph+0.66
Impatience and Irritability+0.61
Contempt+0.56
Contemplation-0.75
Affection-0.59
Contentment-0.56
VoiceNet dimension correlation
Pitch Range+0.96
Arousal+0.94
Metallic Character+0.93
Monologue Style-0.89
Whisper-Talk Style-0.88
Warmth-0.86
Listen — 3 samples per dose
25%measured target: 0.00 · 0.54 · 0.03
50%measured target: 0.07 · 0.33 · 0.15
75%measured target: 0.68 · 0.06 · 0.59
100%measured target: 0.71 · 0.68 · 0.65
125% ★measured target: 0.49 · 0.75 · 0.65
Merging Dynamic Arc __high raises the dimension: at the recommended 125% dose the measured Dynamic Arc score moves from a 0.02 baseline to 0.65 (a +0.63 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 (-0.42); across the board genuineness -0.08, burst-blend -0.42, speech quality -0.18. Emotionally it reads as Triumph (r 0.66), Impatience and Irritability (0.61) and Contempt (0.56), and it moves away from Contemplation, Affection and Contentment. On other voice dimensions it co-moves most with Pitch Range (0.96), Arousal (0.94) and Metallic Character (0.93), while pulling down Monologue Style, Whisper-Talk Style and Warmth. 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 Contemplation to keep the dynamic arc shift without the side-effect.
▼ LOW — Dynamic Arc __low
Dose-response
| dose | target | Δtarget | Δgenu | Δblend | Δqual |
|---|
| baseline (0%) | 0.022 | — | — | — | — |
| 25% ★ | 0.000 | -0.02 | -0.01 | +0.10 | +0.02 |
| 50% | 0.016 | -0.01 | +0.00 | +0.11 | -0.01 |
| 75% | 0.030 | +0.01 | -0.05 | +0.06 | +0.06 |
| 100% | 0.060 | +0.04 | +0.00 | +0.07 | +0.00 |
| 125% | 0.077 | +0.05 | +0.04 | -0.02 | -0.17 |
EmoNet emotion correlation
Pain+0.41
Triumph+0.36
Anger+0.36
Awe-0.50
Contemplation-0.49
Interest-0.42
VoiceNet dimension correlation
Playful Style+0.76
Ranting/Angry Style+0.59
Arousal+0.50
Warmth-0.72
Whisper-Talk Style-0.64
Recording Quality-0.57
Listen — 3 samples per dose
25% ★measured target: 0.00 · 0.00 · 0.00
50%measured target: 0.00 · 0.13 · 0.00
75%measured target: 0.01 · 0.00 · 0.00
100%measured target: 0.00 · 0.11 · 0.00
125%measured target: 0.11 · 0.00 · 0.07
Merging Dynamic Arc __low dials back the dimension: at the recommended 25% dose the measured Dynamic Arc score moves from a 0.02 baseline to 0.00 (a -0.02 shift). The dose-response is noisy and non-monotonic — the merge is an unreliable knob here. That reach is nearly free — the biggest hit is to genuineness (-0.01); across the board genuineness -0.01, burst-blend +0.10, speech quality +0.02. Emotionally it reads as Pain (r 0.41), Triumph (0.36) and Anger (0.36), and it moves away from Awe, Contemplation and Interest. On other voice dimensions it co-moves most with Playful Style (0.76), Ranting/Angry Style (0.59) and Arousal (0.50), while pulling down Warmth, Whisper-Talk Style and Recording Quality. Practical call: merge __low at a light 25–50%; expect Pain and Playful Style to rise and genuineness to fall. If you don’t want Pain bleeding in, offset it with a prompt cue toward Awe to keep the dynamic arc shift without the side-effect.