The merged parent used in these specialist comparisons is now released as Humaneness Voice Large 1.6 Alpha: 1.5 + final RealSpeech rank 4, merged once. Apply only the selected new bucket adapter. Do not reload either ancestor.
1.5 Original model · 1.6 merged model · Complete context manuals for humans and agents
Read 1.5 context manual · Read 1.6 + adapter context manual. Exact prompts, CFG, adapter scales and failed gates are retained. New model/documentation contributions: CC BY 4.0; existing upstream and gated specialist terms remain applicable.
The original grid used generic GENERAL/SCRIPT directions at CFG1. It did not systematically use target-specific manual winners. The new study freezes the actual rank-4 manual wording, preserves every literal word, seed, reference, adapter and frame budget, then independently varies CFG and adapter strength. The previous recordings and conclusions remain below.
Ranking:100% requested EARS emotion/style head, standardized using frozen training statistics; low-pole targets reverse the sign. Formal/casual contrasts average the desired head and negative opposite head. CE and genuineness are diagnostics. Higher target z means stronger desired delivery, not a human rating. More CFG is not automatically better.
Selection:20 original-text outcomes per configuration select wording, CFG and scale. Fifty additional unseen-text reference pairs test the frozen choices. A recommended adapter must improve against both the same-CFG no-adapter control and the best manual-only setting. Quality gates: per-take WER≤15%, mean WER≤10%, valid/terminated fraction≥90%, paired identity change≥−0.05 over50pairs. Failed outcomes stay visible. Language-cluster bootstrap intervals use only five fixed reference voices; findings across72targets remain exploratory.
Five languages: EN/DE/FR/NL/ES; original seeds1701/1702; reference187frames/14.96seconds. CFG1/1.25/1.5/2/2.5/3/4/5 × bucket strength0/0.5/0.75/1. CFG4/5 are stress conditions, not presumed improvements. No-reference use of these bucket recommendations has not been validated.
Taxonomy:S_MONO means Monologue, S_TECH means Teacher/Didactic, VOLT means Volatility. The two corrected style recipes and17dimension/contrast recipes are new searches, not historical manual winners.53other targets reuse directly applicable manual recipes.
Best tested agent route: No new configuration passes the held-out gates; retain standard fallback and inspect audio. Bucket scale 0.0, CFG 1.0.
| Held-out condition | n | Target z↑ | Raw head | CE↑ | Genuine↑ | Voice↑ | WER↓ | Valid | QA |
|---|---|---|---|---|---|---|---|---|---|
| standard | 50 | -0.083 | 0.149 | 6.283 | 4.377 | 0.577 | 0.011 | 1.000 | pass |
| manual | 50 | 1.628 | 1.053 | 6.096 | 4.062 | 0.506 | 0.005 | 1.000 | fail |
| adapter | 50 | 0.432 | 0.421 | 6.271 | 4.325 | 0.549 | 0.005 | 1.000 | pass |
| matched_no_adapter | 50 | 0.341 | 0.373 | 6.258 | 4.281 | 0.571 | 0.005 | 1.000 | pass |
Incremental adapter gain versus same-CFG control: 0.091 z;95% cluster interval [-0.039072271920715586, 0.2362597442993796]. Versus best manual-only: -1.196 z. Manual-only versus standard: 1.711 z.
| Scale | CFG | Target z | CE | Genuine | WER | Provisional QA |
|---|---|---|---|---|---|---|
| 0.0 | 1.0 | 0.482 | 6.203 | 4.319 | 0.003 | False |
| 0.5 | 1.0 | 0.459 | 6.207 | 4.351 | 0.000 | False |
| 0.75 | 1.0 | 0.404 | 6.127 | 4.281 | 0.003 | True |
| 1.0 | 1.0 | 0.525 | 6.182 | 4.318 | 0.008 | True |
| 0.0 | 1.25 | 0.778 | 6.165 | 4.244 | 0.025 | False |
| 0.5 | 1.25 | 0.636 | 6.176 | 4.282 | 0.003 | True |
| 0.75 | 1.25 | 0.536 | 6.188 | 4.277 | 0.008 | False |
| 1.0 | 1.25 | 0.730 | 6.198 | 4.345 | 0.000 | False |
| 0.0 | 1.5 | 0.708 | 6.173 | 4.231 | 0.006 | False |
| 0.5 | 1.5 | 0.687 | 6.223 | 4.278 | 0.000 | False |
| 0.75 | 1.5 | 0.871 | 6.181 | 4.321 | 0.011 | False |
| 1.0 | 1.5 | 0.778 | 6.191 | 4.228 | 0.000 | False |
| 0.0 | 2.0 | 1.224 | 6.127 | 4.217 | 0.014 | False |
| 0.5 | 2.0 | 1.037 | 6.118 | 4.210 | 0.006 | False |
| 0.75 | 2.0 | 1.236 | 6.162 | 4.169 | 0.003 | False |
| 1.0 | 2.0 | 1.017 | 6.166 | 4.223 | 0.000 | False |
| 0.0 | 2.5 | 1.731 | 6.076 | 4.076 | 0.011 | False |
| 0.5 | 2.5 | 1.379 | 6.072 | 4.133 | 0.000 | False |
| 0.75 | 2.5 | 1.648 | 6.114 | 4.134 | 0.000 | False |
| 1.0 | 2.5 | 1.654 | 6.063 | 4.097 | 0.000 | False |
| 0.0 | 3.0 | 2.213 | 5.989 | 4.074 | 0.003 | False |
| 0.5 | 3.0 | 1.931 | 6.060 | 4.100 | 0.017 | False |
| 0.75 | 3.0 | 2.015 | 6.015 | 4.070 | 0.000 | False |
| 1.0 | 3.0 | 2.166 | 5.996 | 4.036 | 0.003 | False |
| 0.0 | 4.0 | 2.173 | 5.860 | 4.001 | 0.003 | False |
| 0.5 | 4.0 | 2.473 | 5.819 | 3.973 | 0.012 | False |
| 0.75 | 4.0 | 2.387 | 5.828 | 3.958 | 0.008 | False |
| 1.0 | 4.0 | 2.477 | 5.902 | 3.956 | 0.006 | False |
| 0.0 | 5.0 | 2.150 | 5.782 | 3.982 | 0.027 | False |
| 0.5 | 5.0 | 2.435 | 5.707 | 4.009 | 0.017 | False |
| 0.75 | 5.0 | 2.660 | 5.686 | 3.983 | 0.005 | False |
| 1.0 | 5.0 | 2.823 | 5.748 | 3.928 | 0.011 | False |
direct_manual_recipe
{
"id": "manual-f9f56f50b9e3",
"general": "Employ an accelerating, breathless quality, treating the triumph as an unstoppable, joyous momentum that carries the speech forward.",
"inline": [
"Start with a rapid, energized delivery, implying the immediate aftermath of a massive effort.",
"Increase the pace and vocal projection steadily, maintaining high forward momentum through the core of the thought.",
"Let the energy peak on the last word, sustaining a bright, elated projection that feels like an outward surge."
],
"cfg": 3,
"recipe": "P06",
"strategy": "local-Gemma-mutation",
"timing_syntax": "canonical",
"canonical_labels": true,
"neutral_event": "keep",
"parent": "cfg-3",
"origin": "exact_frozen_manual_recipe"
}Exact machine-readable comparison · Agent tool recommendations
Intense Triumph. Base: 1.5 Alpha + merged real-speech rank4; new adapter: rank32. Every row is a matched seed/text/reference test. Scale zero disables the new adapter; all other scales are applied absolutely, never compounded.
Adapter training data: {"en": 22, "nl": 5, "de": 5, "es": 1, "fr": 2}. All available accepted languages were pooled, globally shuffled, for three epochs. Peak LR 2e-05 at the end of the first-epoch warmup, then cosine decay. All original optimizer/RNG checkpoints remain saved.
Taxonomy: S_MONO is monologue, not monotone; VOLT is volatility, not volume. No claim is made that a classifier score alone proves the audible performance.
| Condition | n | Target score ↑ | Content Enjoyment ↑ | Genuineness ↑ | WER ↓ |
|---|---|---|---|---|---|
| Merged base, no bucket adapter | 20 | 0.479 ± 0.180 | 6.225 ± 0.091 | 4.292 ± 0.202 | 0.53% ± 2.35 |
| Rank32 × 0.5 | 20 | 0.474 ± 0.178 | 6.185 ± 0.138 | 4.320 ± 0.238 | 0.26% ± 1.18 |
| Rank32 × 0.75 | 20 | 0.486 ± 0.176 | 6.236 ± 0.103 | 4.303 ± 0.201 | 0.00% ± 0.00 |
| Rank32 × 1.0 | 20 | 0.476 ± 0.170 | 6.171 ± 0.165 | 4.324 ± 0.240 | 0.00% ± 0.00 |
Fixed speaker reference:
| Condition | n | Target score ↑ | Content Enjoyment ↑ | Genuineness ↑ | WER ↓ |
|---|---|---|---|---|---|
| Merged base, no bucket adapter | 4 | 0.423 ± 0.013 | 6.305 ± 0.021 | 4.225 ± 0.179 | 0.00% ± 0.00 |
| Rank32 × 0.5 | 4 | 0.426 ± 0.029 | 6.271 ± 0.101 | 4.176 ± 0.235 | 0.00% ± 0.00 |
| Rank32 × 0.75 | 4 | 0.423 ± 0.005 | 6.255 ± 0.097 | 4.146 ± 0.119 | 0.00% ± 0.00 |
| Rank32 × 1.0 | 4 | 0.399 ± 0.004 | 6.245 ± 0.136 | 4.117 ± 0.350 | 0.00% ± 0.00 |
| Prompt / seed | Merged base · no new LoRA | Rank32 × 0.5 | Rank32 × 0.75 | Rank32 × 1.0 |
|---|---|---|---|---|
| Seed 1701 The door is still open. I put your coat on the chair. Tell me when you are ready. Exact GENERAL / SCRIPT promptGENERAL: One adult speaker in a clean recording. Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] The door is still open. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.8 seconds duration] I put your coat on the chair. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Tell me when you are ready. | Target 0.422 · CE 6.30 · Genuine 4.08 · WER 0.0% | Target 0.408 · CE 6.25 · Genuine 3.99 · WER 0.0% | Target 0.424 · CE 6.32 · Genuine 4.08 · WER 0.0% | Target 0.396 · CE 6.28 · Genuine 3.95 · WER 0.0% |
| Seed 1702 The door is still open. I put your coat on the chair. Tell me when you are ready. Exact GENERAL / SCRIPT promptGENERAL: One adult speaker in a clean recording. Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] The door is still open. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.8 seconds duration] I put your coat on the chair. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Tell me when you are ready. | Target 0.407 · CE 6.32 · Genuine 4.13 · WER 0.0% | Target 0.445 · CE 6.17 · Genuine 3.96 · WER 0.0% | Target 0.421 · CE 6.13 · Genuine 4.04 · WER 0.0% | Target 0.398 · CE 6.06 · Genuine 3.83 · WER 0.0% |
| Seed 1701 I found the little box. It was beside the old window. We can look inside together. Exact GENERAL / SCRIPT promptGENERAL: One adult speaker in a clean recording. Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] I found the little box. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.8 seconds duration] It was beside the old window. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] We can look inside together. | Target 0.438 · CE 6.32 · Genuine 4.48 · WER 0.0% | Target 0.456 · CE 6.41 · Genuine 4.35 · WER 0.0% | Target 0.419 · CE 6.34 · Genuine 4.31 · WER 0.0% | Target 0.405 · CE 6.39 · Genuine 4.62 · WER 0.0% |
| Seed 1702 I found the little box. It was beside the old window. We can look inside together. Exact GENERAL / SCRIPT promptGENERAL: One adult speaker in a clean recording. Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] I found the little box. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.8 seconds duration] It was beside the old window. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] We can look inside together. | Target 0.425 · CE 6.28 · Genuine 4.21 · WER 0.0% | Target 0.395 · CE 6.25 · Genuine 4.41 · WER 0.0% | Target 0.430 · CE 6.24 · Genuine 4.16 · WER 0.0% | Target 0.397 · CE 6.25 · Genuine 4.07 · WER 0.0% |
Fixed speaker reference:
| Condition | n | Target score ↑ | Content Enjoyment ↑ | Genuineness ↑ | WER ↓ |
|---|---|---|---|---|---|
| Merged base, no bucket adapter | 4 | 0.726 ± 0.309 | 6.303 ± 0.084 | 4.267 ± 0.217 | 0.00% ± 0.00 |
| Rank32 × 0.5 | 4 | 0.716 ± 0.314 | 6.275 ± 0.073 | 4.322 ± 0.164 | 0.00% ± 0.00 |
| Rank32 × 0.75 | 4 | 0.770 ± 0.244 | 6.316 ± 0.112 | 4.277 ± 0.144 | 0.00% ± 0.00 |
| Rank32 × 1.0 | 4 | 0.681 ± 0.310 | 6.241 ± 0.189 | 4.478 ± 0.115 | 0.00% ± 0.00 |
| Prompt / seed | Merged base · no new LoRA | Rank32 × 0.5 | Rank32 × 0.75 | Rank32 × 1.0 |
|---|---|---|---|---|
| Seed 1701 Die Tür ist noch offen. Ich habe deinen Mantel auf den Stuhl gelegt. Sag mir, wenn du bereit bist. Exact GENERAL / SCRIPT promptGENERAL: One adult speaker in a clean recording. Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Die Tür ist noch offen. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.8 seconds duration] Ich habe deinen Mantel auf den Stuhl gelegt. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Sag mir, wenn du bereit bist. | Target 0.452 · CE 6.23 · Genuine 4.22 · WER 0.0% | Target 0.463 · CE 6.25 · Genuine 4.48 · WER 0.0% | Target 0.626 · CE 6.36 · Genuine 4.43 · WER 0.0% | Target 0.404 · CE 6.28 · Genuine 4.52 · WER 0.0% |
| Seed 1702 Die Tür ist noch offen. Ich habe deinen Mantel auf den Stuhl gelegt. Sag mir, wenn du bereit bist. Exact GENERAL / SCRIPT promptGENERAL: One adult speaker in a clean recording. Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Die Tür ist noch offen. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.8 seconds duration] Ich habe deinen Mantel auf den Stuhl gelegt. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Sag mir, wenn du bereit bist. | Target 0.465 · CE 6.25 · Genuine 4.57 · WER 0.0% | Target 0.425 · CE 6.20 · Genuine 4.44 · WER 0.0% | Target 0.500 · CE 6.16 · Genuine 4.34 · WER 0.0% | Target 0.422 · CE 5.97 · Genuine 4.62 · WER 0.0% |
| Seed 1701 Ich habe die kleine Schachtel gefunden. Sie stand neben dem alten Fenster. Wir können gemeinsam hineinschauen. Exact GENERAL / SCRIPT promptGENERAL: One adult speaker in a clean recording. Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Ich habe die kleine Schachtel gefunden. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.8 seconds duration] Sie stand neben dem alten Fenster. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Wir können gemeinsam hineinschauen. | Target 0.999 · CE 6.41 · Genuine 4.06 · WER 0.0% | Target 0.989 · CE 6.37 · Genuine 4.17 · WER 0.0% | Target 0.968 · CE 6.42 · Genuine 4.10 · WER 0.0% | Target 0.926 · CE 6.40 · Genuine 4.40 · WER 0.0% |
| Seed 1702 Ich habe die kleine Schachtel gefunden. Sie stand neben dem alten Fenster. Wir können gemeinsam hineinschauen. Exact GENERAL / SCRIPT promptGENERAL: One adult speaker in a clean recording. Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Ich habe die kleine Schachtel gefunden. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.8 seconds duration] Sie stand neben dem alten Fenster. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Wir können gemeinsam hineinschauen. | Target 0.987 · CE 6.32 · Genuine 4.22 · WER 0.0% | Target 0.985 · CE 6.28 · Genuine 4.19 · WER 0.0% | Target 0.984 · CE 6.33 · Genuine 4.23 · WER 0.0% | Target 0.973 · CE 6.32 · Genuine 4.37 · WER 0.0% |
Fixed speaker reference:
| Condition | n | Target score ↑ | Content Enjoyment ↑ | Genuineness ↑ | WER ↓ |
|---|---|---|---|---|---|
| Merged base, no bucket adapter | 4 | 0.403 ± 0.064 | 6.223 ± 0.073 | 4.423 ± 0.295 | 2.63% ± 5.26 |
| Rank32 × 0.5 | 4 | 0.409 ± 0.052 | 6.120 ± 0.043 | 4.518 ± 0.227 | 1.32% ± 2.63 |
| Rank32 × 0.75 | 4 | 0.392 ± 0.019 | 6.269 ± 0.058 | 4.479 ± 0.162 | 0.00% ± 0.00 |
| Rank32 × 1.0 | 4 | 0.459 ± 0.111 | 6.249 ± 0.078 | 4.468 ± 0.111 | 0.00% ± 0.00 |
| Prompt / seed | Merged base · no new LoRA | Rank32 × 0.5 | Rank32 × 0.75 | Rank32 × 1.0 |
|---|---|---|---|---|
| Seed 1701 La porte est encore ouverte. J’ai posé ton manteau sur la chaise. Dis-moi quand tu es prêt. Exact GENERAL / SCRIPT promptGENERAL: One adult speaker in a clean recording. Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] La porte est encore ouverte. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.8 seconds duration] J’ai posé ton manteau sur la chaise. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Dis-moi quand tu es prêt. | Target 0.376 · CE 6.17 · Genuine 4.60 · WER 0.0% | Target 0.380 · CE 6.09 · Genuine 4.41 · WER 0.0% | Target 0.381 · CE 6.24 · Genuine 4.44 · WER 0.0% | Target 0.352 · CE 6.25 · Genuine 4.63 · WER 0.0% |
| Seed 1702 La porte est encore ouverte. J’ai posé ton manteau sur la chaise. Dis-moi quand tu es prêt. Exact GENERAL / SCRIPT promptGENERAL: One adult speaker in a clean recording. Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] La porte est encore ouverte. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.8 seconds duration] J’ai posé ton manteau sur la chaise. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Dis-moi quand tu es prêt. | Target 0.499 · CE 6.20 · Genuine 3.99 · WER 0.0% | Target 0.386 · CE 6.18 · Genuine 4.42 · WER 0.0% | Target 0.372 · CE 6.25 · Genuine 4.27 · WER 0.0% | Target 0.430 · CE 6.25 · Genuine 4.42 · WER 0.0% |
| Seed 1701 J’ai trouvé la petite boîte. Elle était près de la vieille fenêtre. Nous pouvons regarder ensemble ce qu’elle contient. Exact GENERAL / SCRIPT promptGENERAL: One adult speaker in a clean recording. Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] J’ai trouvé la petite boîte. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.8 seconds duration] Elle était près de la vieille fenêtre. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Nous pouvons regarder ensemble ce qu’elle contient. | Target 0.368 · CE 6.19 · Genuine 4.47 · WER 0.0% | Target 0.487 · CE 6.12 · Genuine 4.39 · WER 0.0% | Target 0.401 · CE 6.24 · Genuine 4.56 · WER 0.0% | Target 0.614 · CE 6.35 · Genuine 4.43 · WER 0.0% |
| Seed 1702 J’ai trouvé la petite boîte. Elle était près de la vieille fenêtre. Nous pouvons regarder ensemble ce qu’elle contient. Exact GENERAL / SCRIPT promptGENERAL: One adult speaker in a clean recording. Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] J’ai trouvé la petite boîte. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.8 seconds duration] Elle était près de la vieille fenêtre. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Nous pouvons regarder ensemble ce qu’elle contient. | Target 0.371 · CE 6.33 · Genuine 4.63 · WER 10.5% | Target 0.382 · CE 6.09 · Genuine 4.86 · WER 5.3% | Target 0.414 · CE 6.36 · Genuine 4.65 · WER 0.0% | Target 0.439 · CE 6.16 · Genuine 4.39 · WER 0.0% |
Fixed speaker reference:
| Condition | n | Target score ↑ | Content Enjoyment ↑ | Genuineness ↑ | WER ↓ |
|---|---|---|---|---|---|
| Merged base, no bucket adapter | 4 | 0.437 ± 0.029 | 6.156 ± 0.053 | 4.391 ± 0.061 | 0.00% ± 0.00 |
| Rank32 × 0.5 | 4 | 0.421 ± 0.021 | 6.134 ± 0.133 | 4.272 ± 0.083 | 0.00% ± 0.00 |
| Rank32 × 0.75 | 4 | 0.422 ± 0.026 | 6.187 ± 0.054 | 4.313 ± 0.103 | 0.00% ± 0.00 |
| Rank32 × 1.0 | 4 | 0.430 ± 0.022 | 6.128 ± 0.083 | 4.311 ± 0.190 | 0.00% ± 0.00 |
| Prompt / seed | Merged base · no new LoRA | Rank32 × 0.5 | Rank32 × 0.75 | Rank32 × 1.0 |
|---|---|---|---|---|
| Seed 1701 De deur staat nog open. Ik heb je jas op de stoel gelegd. Zeg maar wanneer je klaar bent. Exact GENERAL / SCRIPT promptGENERAL: One adult speaker in a clean recording. Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] De deur staat nog open. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.8 seconds duration] Ik heb je jas op de stoel gelegd. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Zeg maar wanneer je klaar bent. | Target 0.421 · CE 6.10 · Genuine 4.33 · WER 0.0% | Target 0.415 · CE 6.14 · Genuine 4.27 · WER 0.0% | Target 0.455 · CE 6.21 · Genuine 4.44 · WER 0.0% | Target 0.461 · CE 6.18 · Genuine 4.33 · WER 0.0% |
| Seed 1702 De deur staat nog open. Ik heb je jas op de stoel gelegd. Zeg maar wanneer je klaar bent. Exact GENERAL / SCRIPT promptGENERAL: One adult speaker in a clean recording. Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] De deur staat nog open. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.8 seconds duration] Ik heb je jas op de stoel gelegd. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Zeg maar wanneer je klaar bent. | Target 0.429 · CE 6.13 · Genuine 4.41 · WER 0.0% | Target 0.395 · CE 5.97 · Genuine 4.16 · WER 0.0% | Target 0.393 · CE 6.11 · Genuine 4.24 · WER 0.0% | Target 0.414 · CE 6.01 · Genuine 4.04 · WER 0.0% |
| Seed 1701 Ik heb het kleine doosje gevonden. Het stond naast het oude raam. We kunnen er samen in kijken. Exact GENERAL / SCRIPT promptGENERAL: One adult speaker in a clean recording. Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Ik heb het kleine doosje gevonden. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.8 seconds duration] Het stond naast het oude raam. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] We kunnen er samen in kijken. | Target 0.480 · CE 6.22 · Genuine 4.36 · WER 0.0% | Target 0.445 · CE 6.29 · Genuine 4.32 · WER 0.0% | Target 0.421 · CE 6.20 · Genuine 4.34 · WER 0.0% | Target 0.424 · CE 6.12 · Genuine 4.42 · WER 0.0% |
| Seed 1702 Ik heb het kleine doosje gevonden. Het stond naast het oude raam. We kunnen er samen in kijken. Exact GENERAL / SCRIPT promptGENERAL: One adult speaker in a clean recording. Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Ik heb het kleine doosje gevonden. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.8 seconds duration] Het stond naast het oude raam. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] We kunnen er samen in kijken. | Target 0.417 · CE 6.17 · Genuine 4.47 · WER 0.0% | Target 0.429 · CE 6.13 · Genuine 4.34 · WER 0.0% | Target 0.417 · CE 6.24 · Genuine 4.22 · WER 0.0% | Target 0.420 · CE 6.20 · Genuine 4.46 · WER 0.0% |
Fixed speaker reference:
| Condition | n | Target score ↑ | Content Enjoyment ↑ | Genuineness ↑ | WER ↓ |
|---|---|---|---|---|---|
| Merged base, no bucket adapter | 4 | 0.407 ± 0.048 | 6.140 ± 0.064 | 4.154 ± 0.148 | 0.00% ± 0.00 |
| Rank32 × 0.5 | 4 | 0.396 ± 0.025 | 6.124 ± 0.227 | 4.313 ± 0.365 | 0.00% ± 0.00 |
| Rank32 × 0.75 | 4 | 0.425 ± 0.034 | 6.154 ± 0.129 | 4.299 ± 0.330 | 0.00% ± 0.00 |
| Rank32 × 1.0 | 4 | 0.411 ± 0.039 | 5.991 ± 0.196 | 4.247 ± 0.234 | 0.00% ± 0.00 |
| Prompt / seed | Merged base · no new LoRA | Rank32 × 0.5 | Rank32 × 0.75 | Rank32 × 1.0 |
|---|---|---|---|---|
| Seed 1701 La puerta sigue abierta. He dejado tu abrigo sobre la silla. Dime cuándo estás listo. Exact GENERAL / SCRIPT promptGENERAL: One adult speaker in a clean recording. Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] La puerta sigue abierta. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.8 seconds duration] He dejado tu abrigo sobre la silla. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Dime cuándo estás listo. | Target 0.429 · CE 6.08 · Genuine 3.99 · WER 0.0% | Target 0.432 · CE 6.17 · Genuine 3.95 · WER 0.0% | Target 0.386 · CE 6.28 · Genuine 4.49 · WER 0.0% | Target 0.377 · CE 5.89 · Genuine 4.26 · WER 0.0% |
| Seed 1702 La puerta sigue abierta. He dejado tu abrigo sobre la silla. Dime cuándo estás listo. Exact GENERAL / SCRIPT promptGENERAL: One adult speaker in a clean recording. Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] La puerta sigue abierta. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.8 seconds duration] He dejado tu abrigo sobre la silla. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Dime cuándo estás listo. | Target 0.463 · CE 6.16 · Genuine 4.15 · WER 0.0% | Target 0.384 · CE 5.81 · Genuine 4.05 · WER 0.0% | Target 0.416 · CE 5.97 · Genuine 3.83 · WER 0.0% | Target 0.380 · CE 5.77 · Genuine 3.92 · WER 0.0% |
| Seed 1701 He encontrado la cajita. Estaba junto a la ventana vieja. Podemos mirar juntos lo que hay dentro. Exact GENERAL / SCRIPT promptGENERAL: One adult speaker in a clean recording. Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] He encontrado la cajita. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.8 seconds duration] Estaba junto a la ventana vieja. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Podemos mirar juntos lo que hay dentro. | Target 0.377 · CE 6.10 · Genuine 4.12 · WER 0.0% | Target 0.374 · CE 6.35 · Genuine 4.61 · WER 0.0% | Target 0.469 · CE 6.21 · Genuine 4.31 · WER 0.0% | Target 0.459 · CE 6.18 · Genuine 4.36 · WER 0.0% |
| Seed 1702 He encontrado la cajita. Estaba junto a la ventana vieja. Podemos mirar juntos lo que hay dentro. Exact GENERAL / SCRIPT promptGENERAL: One adult speaker in a clean recording. Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] He encontrado la cajita. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.8 seconds duration] Estaba junto a la ventana vieja. (Intense Triumph. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Podemos mirar juntos lo que hay dentro. | Target 0.358 · CE 6.22 · Genuine 4.35 · WER 0.0% | Target 0.396 · CE 6.17 · Genuine 4.65 · WER 0.0% | Target 0.427 · CE 6.16 · Genuine 4.56 · WER 0.0% | Target 0.427 · CE 6.13 · Genuine 4.46 · WER 0.0% |