← all dimensions · markdown (agent skill) · manual home
Voice Age VoiceNet dimension guide
Speaker Identitycode vn_AGEVslot 52
This dimension measures the perceived biological age and vocal wear 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.
Voice Age controls the perceived biological age and vocal wear of the speaker. The __high LoRA shifts it +0.21 at 100% (worst cost: burst-blend -0.18); the __low LoRA shifts it -0.51 at 125% (worst cost: speech quality -0.28). The __low direction is the stronger, more reliable knob here.
▲ HIGH — Voice Age __high
Dose-response
| dose | target | Δtarget | Δgenu | Δblend | Δqual |
|---|
| baseline (0%) | 0.643 | — | — | — | — |
| 25% | 0.612 | -0.03 | -0.01 | -0.01 | +0.02 |
| 50% | 0.589 | -0.05 | -0.04 | -0.07 | +0.04 |
| 75% | 0.702 | +0.06 | -0.07 | -0.14 | +0.11 |
| 100% ★ | 0.848 | +0.21 | -0.09 | -0.18 | +0.15 |
| 125% | 0.797 | +0.15 | -0.11 | -0.29 | +0.21 |
EmoNet emotion correlation
Concentration+0.38
Emotional Numbness+0.37
Contemplation+0.35
Impatience and Irritability-0.39
Confusion-0.36
Pain-0.34
VoiceNet dimension correlation
Whisper-Talk Style+0.77
Fullness+0.77
Warmth+0.69
Playful Style-0.53
Dynamic Arc-0.47
Casual Style-0.43
Listen — 3 samples per dose
25%measured target: 0.63 · 0.62 · 0.85
50%measured target: 0.67 · 0.69 · 0.23
75%measured target: 0.84 · 0.72 · 0.68
100% ★measured target: 0.92 · 0.90 · 0.82
125%measured target: 0.92 · 0.72 · 0.84
Merging Voice Age __high pushes the dimension: at the recommended 100% dose the measured Voice Age score moves from a 0.64 baseline to 0.85 (a +0.21 shift). The dose-response is mostly ordered but noisy between doses. That reach is a real cost — the biggest hit is to burst-blend (-0.18); across the board genuineness -0.09, burst-blend -0.18, speech quality +0.15. Emotionally it reads as Concentration (r 0.38), Emotional Numbness (0.37) and Contemplation (0.35), and it moves away from Impatience and Irritability, Confusion and Pain. On other voice dimensions it co-moves most with Whisper-Talk Style (0.77), Fullness (0.77) and Warmth (0.69), while pulling down Playful Style, Dynamic Arc and Casual Style. Practical call: merge __high at 75–100%; expect Concentration and Whisper-Talk Style to rise and burst-blend to fall. If you don’t want Concentration bleeding in, offset it with a prompt cue toward Impatience and Irritability to keep the voice age shift without the side-effect.
▼ LOW — Voice Age __low
Dose-response
| dose | target | Δtarget | Δgenu | Δblend | Δqual |
|---|
| baseline (0%) | 0.643 | — | — | — | — |
| 25% | 0.542 | -0.10 | +0.02 | -0.06 | +0.03 |
| 50% | 0.468 | -0.17 | -0.01 | -0.17 | -0.06 |
| 75% | 0.348 | -0.29 | -0.01 | -0.20 | -0.11 |
| 100% | 0.163 | -0.48 | +0.09 | -0.08 | -0.34 |
| 125% ★ | 0.132 | -0.51 | +0.00 | -0.19 | -0.28 |
EmoNet emotion correlation
Interest+0.75
Affection+0.66
Longing+0.66
Pain-0.58
Amusement-0.52
Astonishment/Surprise-0.45
VoiceNet dimension correlation
Fullness+0.89
Warmth+0.89
Whisper-Talk Style+0.87
Playful Style-0.82
Dynamic Arc-0.75
Ranting/Angry Style-0.67
Listen — 3 samples per dose
25%measured target: 0.53 · 0.60 · 0.58
50%measured target: 0.61 · 0.55 · 0.62
75%measured target: 0.10 · 0.46 · 0.08
100%measured target: 0.25 · 0.32 · 0.04
125% ★measured target: 0.22 · 0.23 · 0.04
Merging Voice Age __low lowers the dimension: at the recommended 125% dose the measured Voice Age score moves from a 0.64 baseline to 0.13 (a -0.51 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 (-0.28); across the board genuineness +0.00, burst-blend -0.19, speech quality -0.28. Emotionally it reads as Interest (r 0.75), Affection (0.66) and Longing (0.66), and it moves away from Pain, Amusement and Astonishment/Surprise. On other voice dimensions it co-moves most with Fullness (0.89), Warmth (0.89) and Whisper-Talk Style (0.87), while pulling down Playful Style, Dynamic Arc and Ranting/Angry Style. Practical call: merge __low at 100–125%; expect Interest and Fullness to rise and speech quality to fall. If you don’t want Interest bleeding in, offset it with a prompt cue toward Pain to keep the voice age shift without the side-effect.