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: Use manual-only configuration; no clear qualifying incremental adapter gain. Bucket scale 0.0, CFG 2.5.
| Held-out condition | n | Target z↑ | Raw head | CE↑ | Genuine↑ | Voice↑ | WER↓ | Valid | QA |
|---|---|---|---|---|---|---|---|---|---|
| standard | 50 | -0.304 | 0.178 | 6.261 | 4.350 | 0.579 | 0.017 | 0.980 | pass |
| manual | 50 | 2.170 | 1.509 | 6.255 | 4.159 | 0.531 | 0.012 | 1.000 | pass |
| adapter | 50 | 2.024 | 1.430 | 6.258 | 4.075 | 0.523 | 0.025 | 0.980 | fail |
| matched_no_adapter | 50 | 2.170 | 1.509 | 6.255 | 4.159 | 0.531 | 0.012 | 1.000 | pass |
Incremental adapter gain versus same-CFG control: -0.147 z;95% cluster interval [-0.3213476036567492, 0.08248047444944137]. Versus best manual-only: -0.147 z. Manual-only versus standard: 2.474 z.
| Scale | CFG | Target z | CE | Genuine | WER | Provisional QA |
|---|---|---|---|---|---|---|
| 0.0 | 1.0 | 1.166 | 6.201 | 4.223 | 0.003 | True |
| 0.5 | 1.0 | 0.961 | 6.244 | 4.251 | 0.000 | True |
| 0.75 | 1.0 | 0.552 | 6.185 | 4.161 | 0.003 | True |
| 1.0 | 1.0 | 0.963 | 6.179 | 4.265 | 0.006 | True |
| 0.0 | 1.25 | 1.206 | 6.240 | 4.206 | 0.007 | True |
| 0.5 | 1.25 | 1.213 | 6.194 | 4.162 | 0.000 | True |
| 0.75 | 1.25 | 1.124 | 6.215 | 4.223 | 0.007 | True |
| 1.0 | 1.25 | 1.239 | 6.178 | 4.198 | 0.003 | True |
| 0.0 | 1.5 | 1.287 | 6.224 | 4.140 | 0.003 | True |
| 0.5 | 1.5 | 1.347 | 6.215 | 4.134 | 0.000 | True |
| 0.75 | 1.5 | 1.502 | 6.212 | 4.180 | 0.006 | False |
| 1.0 | 1.5 | 1.450 | 6.182 | 4.163 | 0.000 | True |
| 0.0 | 2.0 | 2.095 | 6.179 | 4.128 | 0.006 | True |
| 0.5 | 2.0 | 1.965 | 6.215 | 4.084 | 0.000 | True |
| 0.75 | 2.0 | 2.020 | 6.162 | 4.047 | 0.006 | False |
| 1.0 | 2.0 | 1.900 | 6.186 | 4.038 | 0.003 | True |
| 0.0 | 2.5 | 2.193 | 6.197 | 4.055 | 0.013 | True |
| 0.5 | 2.5 | 2.139 | 6.179 | 4.067 | 0.003 | False |
| 0.75 | 2.5 | 2.129 | 6.181 | 4.017 | 0.003 | True |
| 1.0 | 2.5 | 2.147 | 6.192 | 3.995 | 0.022 | True |
| 0.0 | 3.0 | 2.624 | 6.181 | 4.023 | 0.008 | False |
| 0.5 | 3.0 | 2.513 | 6.165 | 4.035 | 0.003 | False |
| 0.75 | 3.0 | 2.599 | 6.133 | 3.957 | 0.018 | False |
| 1.0 | 3.0 | 2.796 | 6.133 | 4.046 | 0.009 | False |
| 0.0 | 4.0 | 3.049 | 6.106 | 3.941 | 0.048 | False |
| 0.5 | 4.0 | 2.896 | 6.134 | 3.955 | 0.016 | False |
| 0.75 | 4.0 | 3.102 | 6.104 | 3.953 | 0.012 | False |
| 1.0 | 4.0 | 2.981 | 6.117 | 3.986 | 0.038 | False |
| 0.0 | 5.0 | 3.267 | 6.093 | 3.954 | 0.066 | False |
| 0.5 | 5.0 | 3.231 | 6.091 | 3.884 | 0.061 | False |
| 0.75 | 5.0 | 3.341 | 6.059 | 3.929 | 0.103 | False |
| 1.0 | 5.0 | 2.925 | 6.043 | 3.892 | 0.084 | False |
direct_manual_recipe
{
"id": "manual-c46a333eb8dd",
"general": "Focus on a deliberate, exaggerated sweetness that hints at underlying, playful challenge.",
"inline": [
"Begin with an overly sweet, almost saccharine inflection, immediately setting up the ironic contrast.",
"Introduce subtle, rhythmic vocal emphasis on the unstressed syllables to create a buoyant, teasing cadence.",
"End with a clear, bright resonance, ensuring the underlying flirtatious challenge is perfectly audible."
],
"cfg": 2.5,
"recipe": "P09",
"strategy": "local-Gemma-mutation",
"timing_syntax": "canonical",
"canonical_labels": true,
"neutral_event": "keep",
"parent": "cfg-2.5",
"origin": "exact_frozen_manual_recipe"
}Exact machine-readable comparison · Agent tool recommendations
Intense Teasing. 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": 39, "de": 28, "fr": 21, "es": 16, "nl": 11}. 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 | 0.452 ± 0.180 | 6.182 ± 0.106 | 4.326 ± 0.248 | 0.28% ± 1.24 |
| Rank32 × 0.5 | 20 | 0.451 ± 0.183 | 6.208 ± 0.114 | 4.391 ± 0.230 | 0.00% ± 0.00 |
| Rank32 × 0.75 | 20 | 0.402 ± 0.142 | 6.158 ± 0.153 | 4.341 ± 0.220 | 0.87% ± 2.15 |
| Rank32 × 1.0 | 20 | 0.419 ± 0.130 | 6.196 ± 0.133 | 4.360 ± 0.236 | 0.53% ± 2.35 |
Fixed speaker reference:
| Condition | n | Target score ↑ | Content Enjoyment ↑ | Genuineness ↑ | WER ↓ |
|---|---|---|---|---|---|
| Merged base, no bucket adapter | 4 | 0.433 ± 0.089 | 6.219 ± 0.133 | 4.181 ± 0.241 | 0.00% ± 0.00 |
| Rank32 × 0.5 | 4 | 0.401 ± 0.028 | 6.275 ± 0.113 | 4.183 ± 0.319 | 0.00% ± 0.00 |
| Rank32 × 0.75 | 4 | 0.400 ± 0.069 | 6.328 ± 0.043 | 4.220 ± 0.187 | 0.00% ± 0.00 |
| Rank32 × 1.0 | 4 | 0.403 ± 0.094 | 6.297 ± 0.159 | 4.209 ± 0.314 | 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 Teasing. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Teasing. 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 Teasing. 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 Teasing. 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 6.31 · Genuine 3.99 · WER 0.0% | Target 0.404 · CE 6.31 · Genuine 4.08 · WER 0.0% | Target 0.371 · CE 6.27 · Genuine 4.03 · WER 0.0% | Target 0.538 · CE 6.09 · Genuine 4.21 · 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 Teasing. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Teasing. 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 Teasing. 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 Teasing. 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 0.357 · CE 6.06 · Genuine 4.01 · WER 0.0% | Target 0.439 · CE 6.11 · Genuine 3.78 · WER 0.0% | Target 0.503 · CE 6.32 · Genuine 4.09 · WER 0.0% | Target 0.321 · CE 6.27 · Genuine 3.77 · 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 Teasing. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Teasing. 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 Teasing. 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 Teasing. 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 0.393 · CE 6.35 · Genuine 4.51 · WER 0.0% | Target 0.386 · CE 6.36 · Genuine 4.48 · WER 0.0% | Target 0.352 · CE 6.36 · Genuine 4.34 · WER 0.0% | Target 0.368 · CE 6.45 · Genuine 4.44 · 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 Teasing. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Teasing. 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 Teasing. 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 Teasing. 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 0.561 · CE 6.16 · Genuine 4.22 · WER 0.0% | Target 0.375 · CE 6.32 · Genuine 4.39 · WER 0.0% | Target 0.374 · CE 6.36 · Genuine 4.41 · WER 0.0% | Target 0.383 · CE 6.38 · Genuine 4.42 · WER 0.0% |
Fixed speaker reference:
| Condition | n | Target score ↑ | Content Enjoyment ↑ | Genuineness ↑ | WER ↓ |
|---|---|---|---|---|---|
| Merged base, no bucket adapter | 4 | 0.512 ± 0.215 | 6.289 ± 0.095 | 4.209 ± 0.371 | 0.00% ± 0.00 |
| Rank32 × 0.5 | 4 | 0.409 ± 0.073 | 6.309 ± 0.087 | 4.462 ± 0.057 | 0.00% ± 0.00 |
| Rank32 × 0.75 | 4 | 0.392 ± 0.048 | 6.253 ± 0.113 | 4.438 ± 0.197 | 0.00% ± 0.00 |
| Rank32 × 1.0 | 4 | 0.411 ± 0.041 | 6.328 ± 0.083 | 4.443 ± 0.105 | 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 Teasing. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Teasing. 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 Teasing. 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 Teasing. 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 0.531 · CE 6.17 · Genuine 4.37 · WER 0.0% | Target 0.514 · CE 6.24 · Genuine 4.41 · WER 0.0% | Target 0.446 · CE 6.09 · Genuine 4.46 · WER 0.0% | Target 0.360 · CE 6.26 · Genuine 4.58 · 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 Teasing. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Teasing. 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 Teasing. 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 Teasing. 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 0.809 · CE 6.33 · Genuine 4.66 · WER 0.0% | Target 0.345 · CE 6.23 · Genuine 4.43 · WER 0.0% | Target 0.330 · CE 6.34 · Genuine 4.71 · WER 0.0% | Target 0.451 · CE 6.25 · Genuine 4.46 · 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 Teasing. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Teasing. 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 Teasing. 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 Teasing. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Wir können gemeinsam hineinschauen. | Target 0.355 · CE 6.27 · Genuine 3.92 · WER 0.0% | Target 0.379 · CE 6.38 · Genuine 4.53 · WER 0.0% | Target 0.389 · CE 6.25 · Genuine 4.27 · WER 0.0% | Target 0.438 · CE 6.40 · Genuine 4.39 · 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 Teasing. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Teasing. 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 Teasing. 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 Teasing. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Wir können gemeinsam hineinschauen. | Target 0.354 · CE 6.39 · Genuine 3.89 · WER 0.0% | Target 0.398 · CE 6.39 · Genuine 4.48 · WER 0.0% | Target 0.401 · CE 6.32 · Genuine 4.32 · WER 0.0% | Target 0.396 · CE 6.40 · Genuine 4.34 · WER 0.0% |
Fixed speaker reference:
| Condition | n | Target score ↑ | Content Enjoyment ↑ | Genuineness ↑ | WER ↓ |
|---|---|---|---|---|---|
| Merged base, no bucket adapter | 4 | 0.335 ± 0.035 | 6.160 ± 0.083 | 4.448 ± 0.148 | 1.39% ± 2.78 |
| Rank32 × 0.5 | 4 | 0.350 ± 0.028 | 6.195 ± 0.059 | 4.506 ± 0.304 | 0.00% ± 0.00 |
| Rank32 × 0.75 | 4 | 0.317 ± 0.023 | 6.100 ± 0.084 | 4.443 ± 0.171 | 0.00% ± 0.00 |
| Rank32 × 1.0 | 4 | 0.324 ± 0.011 | 6.102 ± 0.023 | 4.457 ± 0.127 | 2.63% ± 5.26 |
| 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 Teasing. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Teasing. 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 Teasing. 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 Teasing. 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 0.325 · CE 6.13 · Genuine 4.29 · WER 0.0% | Target 0.332 · CE 6.12 · Genuine 4.51 · WER 0.0% | Target 0.332 · CE 6.05 · Genuine 4.27 · WER 0.0% | Target 0.321 · CE 6.13 · Genuine 4.31 · 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 Teasing. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Teasing. 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 Teasing. 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 Teasing. 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 0.320 · CE 6.18 · Genuine 4.39 · WER 5.6% | Target 0.392 · CE 6.18 · Genuine 4.17 · WER 0.0% | Target 0.337 · CE 6.21 · Genuine 4.49 · WER 0.0% | Target 0.310 · CE 6.08 · Genuine 4.39 · 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 Teasing. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Teasing. 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 Teasing. 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 Teasing. 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 0.309 · CE 6.06 · Genuine 4.46 · WER 0.0% | Target 0.344 · CE 6.24 · Genuine 4.43 · WER 0.0% | Target 0.287 · CE 6.02 · Genuine 4.35 · WER 0.0% | Target 0.337 · CE 6.09 · Genuine 4.56 · 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 Teasing. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Teasing. 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 Teasing. 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 Teasing. 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 0.386 · CE 6.26 · Genuine 4.65 · WER 0.0% | Target 0.334 · CE 6.25 · Genuine 4.91 · WER 0.0% | Target 0.311 · CE 6.11 · Genuine 4.66 · WER 0.0% | Target 0.328 · CE 6.12 · Genuine 4.57 · WER 10.5% |
Fixed speaker reference:
| Condition | n | Target score ↑ | Content Enjoyment ↑ | Genuineness ↑ | WER ↓ |
|---|---|---|---|---|---|
| Merged base, no bucket adapter | 4 | 0.649 ± 0.230 | 6.134 ± 0.091 | 4.280 ± 0.140 | 0.00% ± 0.00 |
| Rank32 × 0.5 | 4 | 0.701 ± 0.280 | 6.102 ± 0.083 | 4.328 ± 0.108 | 0.00% ± 0.00 |
| Rank32 × 0.75 | 4 | 0.550 ± 0.272 | 6.080 ± 0.115 | 4.303 ± 0.100 | 2.70% ± 3.13 |
| Rank32 × 1.0 | 4 | 0.632 ± 0.103 | 6.132 ± 0.089 | 4.286 ± 0.212 | 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 Teasing. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Teasing. 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 Teasing. 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 Teasing. 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 0.952 · CE 6.24 · Genuine 4.18 · WER 0.0% | Target 0.866 · CE 6.15 · Genuine 4.31 · WER 0.0% | Target 0.363 · CE 5.93 · Genuine 4.28 · WER 5.3% | Target 0.747 · CE 6.15 · Genuine 4.36 · 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 Teasing. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Teasing. 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 Teasing. 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 Teasing. 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 0.396 · CE 6.01 · Genuine 4.19 · WER 0.0% | Target 0.363 · CE 5.98 · Genuine 4.29 · WER 0.0% | Target 0.913 · CE 6.10 · Genuine 4.42 · WER 0.0% | Target 0.579 · CE 6.13 · Genuine 4.24 · 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 Teasing. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Teasing. 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 Teasing. 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 Teasing. 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 0.650 · CE 6.13 · Genuine 4.48 · WER 0.0% | Target 0.986 · CE 6.11 · Genuine 4.23 · WER 0.0% | Target 0.603 · CE 6.21 · Genuine 4.18 · WER 0.0% | Target 0.686 · CE 6.01 · Genuine 4.02 · 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 Teasing. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Teasing. 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 Teasing. 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 Teasing. 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 0.599 · CE 6.15 · Genuine 4.26 · WER 0.0% | Target 0.587 · CE 6.17 · Genuine 4.48 · WER 0.0% | Target 0.322 · CE 6.09 · Genuine 4.33 · WER 5.6% | Target 0.517 · CE 6.23 · Genuine 4.52 · WER 0.0% |
Fixed speaker reference:
| Condition | n | Target score ↑ | Content Enjoyment ↑ | Genuineness ↑ | WER ↓ |
|---|---|---|---|---|---|
| Merged base, no bucket adapter | 4 | 0.329 ± 0.034 | 6.108 ± 0.025 | 4.510 ± 0.196 | 0.00% ± 0.00 |
| Rank32 × 0.5 | 4 | 0.392 ± 0.143 | 6.158 ± 0.118 | 4.474 ± 0.168 | 0.00% ± 0.00 |
| Rank32 × 0.75 | 4 | 0.353 ± 0.065 | 6.029 ± 0.166 | 4.300 ± 0.377 | 1.67% ± 3.33 |
| Rank32 × 1.0 | 4 | 0.326 ± 0.021 | 6.122 ± 0.099 | 4.405 ± 0.349 | 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 Teasing. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Teasing. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] La puerta sigue abierta. (Intense Teasing. 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 Teasing. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Dime cuándo estás listo. | Target 0.298 · CE 6.13 · Genuine 4.32 · WER 0.0% | Target 0.604 · CE 6.12 · Genuine 4.33 · WER 0.0% | Target 0.322 · CE 5.78 · Genuine 3.96 · WER 0.0% | Target 0.298 · CE 6.26 · Genuine 4.23 · 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 Teasing. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Teasing. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] La puerta sigue abierta. (Intense Teasing. 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 Teasing. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Dime cuándo estás listo. | Target 0.322 · CE 6.07 · Genuine 4.37 · WER 0.0% | Target 0.296 · CE 6.08 · Genuine 4.34 · WER 0.0% | Target 0.448 · CE 6.11 · Genuine 3.99 · WER 6.7% | Target 0.324 · CE 6.05 · Genuine 4.01 · 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 Teasing. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Teasing. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] He encontrado la cajita. (Intense Teasing. 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 Teasing. 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 0.318 · CE 6.12 · Genuine 4.72 · WER 0.0% | Target 0.324 · CE 6.10 · Genuine 4.56 · WER 0.0% | Target 0.300 · CE 6.15 · Genuine 4.63 · WER 0.0% | Target 0.346 · CE 6.13 · Genuine 4.78 · 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 Teasing. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Teasing. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] He encontrado la cajita. (Intense Teasing. 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 Teasing. 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 6.12 · Genuine 4.63 · WER 0.0% | Target 0.342 · CE 6.33 · Genuine 4.67 · WER 0.0% | Target 0.341 · CE 6.07 · Genuine 4.62 · WER 0.0% | Target 0.338 · CE 6.05 · Genuine 4.60 · WER 0.0% |