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 | 0.458 | 0.526 | 6.250 | 4.315 | 0.583 | 0.005 | 1.000 | pass |
| manual | 50 | 1.361 | 1.017 | 5.874 | 4.498 | 0.526 | 0.005 | 1.000 | fail |
| adapter | 50 | 1.297 | 0.983 | 6.023 | 4.464 | 0.544 | 0.010 | 1.000 | pass |
| matched_no_adapter | 50 | 1.156 | 0.906 | 6.141 | 4.535 | 0.613 | 0.019 | 0.980 | pass |
Incremental adapter gain versus same-CFG control: 0.141 z;95% cluster interval [-0.04078124225805788, 0.2977539176145544]. Versus best manual-only: -0.064 z. Manual-only versus standard: 0.903 z.
| Scale | CFG | Target z | CE | Genuine | WER | Provisional QA |
|---|---|---|---|---|---|---|
| 0.0 | 1.0 | 1.356 | 6.183 | 4.472 | 0.003 | True |
| 0.5 | 1.0 | 1.220 | 6.166 | 4.454 | 0.006 | True |
| 0.75 | 1.0 | 1.349 | 6.138 | 4.432 | 0.003 | True |
| 1.0 | 1.0 | 1.439 | 6.103 | 4.337 | 0.003 | True |
| 0.0 | 1.25 | 1.368 | 6.168 | 4.451 | 0.003 | True |
| 0.5 | 1.25 | 1.678 | 6.145 | 4.443 | 0.005 | True |
| 0.75 | 1.25 | 1.410 | 6.055 | 4.433 | 0.016 | True |
| 1.0 | 1.25 | 1.663 | 6.080 | 4.442 | 0.006 | False |
| 0.0 | 1.5 | 1.574 | 6.133 | 4.494 | 0.003 | False |
| 0.5 | 1.5 | 1.628 | 6.085 | 4.458 | 0.003 | True |
| 0.75 | 1.5 | 1.711 | 6.033 | 4.453 | 0.006 | False |
| 1.0 | 1.5 | 1.632 | 5.996 | 4.411 | 0.003 | False |
| 0.0 | 2.0 | 1.647 | 6.057 | 4.508 | 0.000 | True |
| 0.5 | 2.0 | 2.026 | 5.951 | 4.462 | 0.009 | True |
| 0.75 | 2.0 | 2.041 | 5.917 | 4.382 | 0.009 | True |
| 1.0 | 2.0 | 1.805 | 5.862 | 4.437 | 0.000 | False |
| 0.0 | 2.5 | 1.944 | 5.950 | 4.482 | 0.003 | True |
| 0.5 | 2.5 | 1.961 | 5.858 | 4.486 | 0.009 | True |
| 0.75 | 2.5 | 2.125 | 5.722 | 4.394 | 0.003 | False |
| 1.0 | 2.5 | 2.011 | 5.720 | 4.414 | 0.012 | False |
| 0.0 | 3.0 | 1.945 | 5.881 | 4.500 | 0.006 | True |
| 0.5 | 3.0 | 1.937 | 5.763 | 4.437 | 0.000 | False |
| 0.75 | 3.0 | 1.940 | 5.689 | 4.380 | 0.003 | False |
| 1.0 | 3.0 | 1.916 | 5.630 | 4.349 | 0.014 | False |
| 0.0 | 4.0 | 2.041 | 5.638 | 4.360 | 0.019 | False |
| 0.5 | 4.0 | 2.055 | 5.629 | 4.380 | 0.003 | False |
| 0.75 | 4.0 | 2.086 | 5.502 | 4.352 | 0.013 | False |
| 1.0 | 4.0 | 1.856 | 5.472 | 4.332 | 0.003 | False |
| 0.0 | 5.0 | 2.037 | 5.618 | 4.404 | 0.000 | False |
| 0.5 | 5.0 | 2.134 | 5.468 | 4.306 | 0.008 | False |
| 0.75 | 5.0 | 2.068 | 5.470 | 4.323 | 0.011 | False |
| 1.0 | 5.0 | 2.154 | 5.469 | 4.320 | 0.006 | False |
direct_manual_recipe
{
"id": "manual-bbf5dbc0f2ea",
"general": "Infuse the speech with the delicate texture of a shared secret, making the warmth feel intensely personal and slightly conspiratorial.",
"inline": [
"Begin with a very soft breath support, almost whispering the initial sentiment.",
"Maintain a close, intimate vocal proximity throughout, as if leaning in close to the listener.",
"Let the final sound gently dissolve, leaving a lingering, sweet resonance in the air."
],
"cfg": 3,
"recipe": "P05",
"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 Affection. 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": 56, "nl": 30, "fr": 48, "de": 45, "es": 54}. 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 | 0.701 ± 0.245 | 6.185 ± 0.121 | 4.328 ± 0.196 | 0.00% ± 0.00 |
| Rank32 × 0.5 | 20 | 0.682 ± 0.247 | 6.174 ± 0.160 | 4.291 ± 0.278 | 0.00% ± 0.00 |
| Rank32 × 0.75 | 20 | 0.763 ± 0.226 | 6.218 ± 0.127 | 4.262 ± 0.217 | 0.00% ± 0.00 |
| Rank32 × 1.0 | 20 | 0.721 ± 0.245 | 6.206 ± 0.126 | 4.223 ± 0.205 | 0.28% ± 1.24 |
Fixed speaker reference:
| Condition | n | Target score ↑ | Content Enjoyment ↑ | Genuineness ↑ | WER ↓ |
|---|---|---|---|---|---|
| Merged base, no bucket adapter | 4 | 0.980 ± 0.009 | 6.255 ± 0.059 | 4.205 ± 0.154 | 0.00% ± 0.00 |
| Rank32 × 0.5 | 4 | 0.984 ± 0.011 | 6.313 ± 0.118 | 4.186 ± 0.304 | 0.00% ± 0.00 |
| Rank32 × 0.75 | 4 | 0.980 ± 0.009 | 6.313 ± 0.062 | 4.162 ± 0.211 | 0.00% ± 0.00 |
| Rank32 × 1.0 | 4 | 0.995 ± 0.002 | 6.354 ± 0.057 | 4.372 ± 0.189 | 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 Affection. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Affection. 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 Affection. 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 Affection. 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 0.976 · CE 6.19 · Genuine 4.06 · WER 0.0% | Target 0.971 · CE 6.19 · Genuine 3.85 · WER 0.0% | Target 0.973 · CE 6.30 · Genuine 3.92 · WER 0.0% | Target 0.996 · CE 6.29 · Genuine 4.09 · 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 Affection. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Affection. 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 Affection. 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 Affection. 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 0.993 · CE 6.22 · Genuine 4.16 · WER 0.0% | Target 0.995 · CE 6.24 · Genuine 4.01 · WER 0.0% | Target 0.986 · CE 6.25 · Genuine 4.06 · WER 0.0% | Target 0.996 · CE 6.32 · Genuine 4.46 · 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 Affection. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Affection. 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 Affection. 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 Affection. 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 0.971 · CE 6.30 · Genuine 4.19 · WER 0.0% | Target 0.979 · CE 6.39 · Genuine 4.37 · WER 0.0% | Target 0.972 · CE 6.31 · Genuine 4.33 · WER 0.0% | Target 0.992 · CE 6.38 · Genuine 4.51 · 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 Affection. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Affection. 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 Affection. 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 Affection. 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 0.981 · CE 6.31 · Genuine 4.42 · WER 0.0% | Target 0.991 · CE 6.44 · Genuine 4.51 · WER 0.0% | Target 0.988 · CE 6.40 · Genuine 4.34 · WER 0.0% | Target 0.994 · CE 6.42 · Genuine 4.43 · WER 0.0% |
Fixed speaker reference:
| Condition | n | Target score ↑ | Content Enjoyment ↑ | Genuineness ↑ | WER ↓ |
|---|---|---|---|---|---|
| Merged base, no bucket adapter | 4 | 0.557 ± 0.170 | 6.334 ± 0.101 | 4.394 ± 0.092 | 0.00% ± 0.00 |
| Rank32 × 0.5 | 4 | 0.603 ± 0.167 | 6.249 ± 0.112 | 4.418 ± 0.173 | 0.00% ± 0.00 |
| Rank32 × 0.75 | 4 | 0.717 ± 0.114 | 6.322 ± 0.064 | 4.327 ± 0.220 | 0.00% ± 0.00 |
| Rank32 × 1.0 | 4 | 0.629 ± 0.192 | 6.295 ± 0.027 | 4.336 ± 0.149 | 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 Affection. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Affection. 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 Affection. 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 Affection. 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 0.415 · CE 6.32 · Genuine 4.41 · WER 0.0% | Target 0.761 · CE 6.16 · Genuine 4.61 · WER 0.0% | Target 0.726 · CE 6.26 · Genuine 4.36 · WER 0.0% | Target 0.452 · CE 6.29 · Genuine 4.53 · 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 Affection. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Affection. 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 Affection. 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 Affection. 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 0.412 · CE 6.21 · Genuine 4.33 · WER 0.0% | Target 0.433 · CE 6.15 · Genuine 4.44 · WER 0.0% | Target 0.560 · CE 6.28 · Genuine 4.57 · WER 0.0% | Target 0.500 · CE 6.28 · Genuine 4.24 · 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 Affection. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Affection. 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 Affection. 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 Affection. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Wir können gemeinsam hineinschauen. | Target 0.651 · CE 6.46 · Genuine 4.32 · WER 0.0% | Target 0.487 · CE 6.32 · Genuine 4.43 · WER 0.0% | Target 0.752 · CE 6.40 · Genuine 4.04 · WER 0.0% | Target 0.870 · CE 6.27 · Genuine 4.21 · 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 Affection. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Affection. 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 Affection. 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 Affection. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Wir können gemeinsam hineinschauen. | Target 0.749 · CE 6.35 · Genuine 4.52 · WER 0.0% | Target 0.731 · CE 6.37 · Genuine 4.19 · WER 0.0% | Target 0.832 · CE 6.34 · Genuine 4.34 · WER 0.0% | Target 0.694 · CE 6.33 · Genuine 4.37 · WER 0.0% |
Fixed speaker reference:
| Condition | n | Target score ↑ | Content Enjoyment ↑ | Genuineness ↑ | WER ↓ |
|---|---|---|---|---|---|
| Merged base, no bucket adapter | 4 | 0.878 ± 0.115 | 6.168 ± 0.081 | 4.481 ± 0.190 | 0.00% ± 0.00 |
| Rank32 × 0.5 | 4 | 0.834 ± 0.209 | 6.134 ± 0.115 | 4.404 ± 0.261 | 0.00% ± 0.00 |
| Rank32 × 0.75 | 4 | 0.938 ± 0.051 | 6.124 ± 0.163 | 4.189 ± 0.217 | 0.00% ± 0.00 |
| Rank32 × 1.0 | 4 | 0.842 ± 0.139 | 6.132 ± 0.012 | 3.996 ± 0.124 | 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 Affection. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Affection. 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 Affection. 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 Affection. 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 0.794 · CE 6.22 · Genuine 4.22 · WER 0.0% | Target 0.535 · CE 6.01 · Genuine 4.03 · WER 0.0% | Target 0.881 · CE 5.90 · Genuine 3.94 · WER 0.0% | Target 0.726 · CE 6.14 · Genuine 3.88 · 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 Affection. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Affection. 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 Affection. 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 Affection. 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 0.764 · CE 6.05 · Genuine 4.66 · WER 0.0% | Target 0.847 · CE 6.28 · Genuine 4.42 · WER 0.0% | Target 0.910 · CE 6.13 · Genuine 4.08 · WER 0.0% | Target 0.718 · CE 6.13 · Genuine 3.92 · 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 Affection. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Affection. 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 Affection. 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 Affection. 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 0.968 · CE 6.17 · Genuine 4.46 · WER 0.0% | Target 0.954 · CE 6.15 · Genuine 4.56 · WER 0.0% | Target 0.970 · CE 6.21 · Genuine 4.42 · WER 0.0% | Target 0.939 · CE 6.14 · Genuine 4.03 · 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 Affection. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Affection. 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 Affection. 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 Affection. 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 0.986 · CE 6.23 · Genuine 4.58 · WER 0.0% | Target 1.000 · CE 6.09 · Genuine 4.61 · WER 0.0% | Target 0.990 · CE 6.27 · Genuine 4.31 · WER 0.0% | Target 0.984 · CE 6.12 · Genuine 4.16 · WER 0.0% |
Fixed speaker reference:
| Condition | n | Target score ↑ | Content Enjoyment ↑ | Genuineness ↑ | WER ↓ |
|---|---|---|---|---|---|
| Merged base, no bucket adapter | 4 | 0.378 ± 0.035 | 6.094 ± 0.058 | 4.178 ± 0.152 | 0.00% ± 0.00 |
| Rank32 × 0.5 | 4 | 0.372 ± 0.021 | 6.080 ± 0.123 | 4.200 ± 0.105 | 0.00% ± 0.00 |
| Rank32 × 0.75 | 4 | 0.401 ± 0.031 | 6.130 ± 0.050 | 4.200 ± 0.137 | 0.00% ± 0.00 |
| Rank32 × 1.0 | 4 | 0.365 ± 0.013 | 6.156 ± 0.105 | 4.299 ± 0.152 | 1.39% ± 2.78 |
| 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 Affection. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Affection. 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 Affection. 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 Affection. 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 0.331 · CE 6.16 · Genuine 4.04 · WER 0.0% | Target 0.348 · CE 5.91 · Genuine 4.30 · WER 0.0% | Target 0.366 · CE 6.13 · Genuine 4.27 · WER 0.0% | Target 0.354 · CE 6.16 · Genuine 4.43 · 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 Affection. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Affection. 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 Affection. 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 Affection. 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 0.377 · CE 6.08 · Genuine 4.32 · WER 0.0% | Target 0.373 · CE 6.10 · Genuine 4.06 · WER 0.0% | Target 0.392 · CE 6.08 · Genuine 4.34 · WER 0.0% | Target 0.354 · CE 6.01 · Genuine 4.09 · 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 Affection. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Affection. 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 Affection. 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 Affection. 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 0.414 · CE 6.12 · Genuine 4.30 · WER 0.0% | Target 0.400 · CE 6.10 · Genuine 4.20 · WER 0.0% | Target 0.440 · CE 6.20 · Genuine 4.16 · WER 0.0% | Target 0.376 · CE 6.23 · Genuine 4.38 · WER 5.6% |
| 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 Affection. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Affection. 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 Affection. 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 Affection. 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 0.390 · CE 6.02 · Genuine 4.05 · WER 0.0% | Target 0.367 · CE 6.21 · Genuine 4.24 · WER 0.0% | Target 0.405 · CE 6.12 · Genuine 4.03 · WER 0.0% | Target 0.377 · CE 6.23 · Genuine 4.29 · WER 0.0% |
Fixed speaker reference:
| Condition | n | Target score ↑ | Content Enjoyment ↑ | Genuineness ↑ | WER ↓ |
|---|---|---|---|---|---|
| Merged base, no bucket adapter | 4 | 0.711 ± 0.152 | 6.076 ± 0.071 | 4.385 ± 0.248 | 0.00% ± 0.00 |
| Rank32 × 0.5 | 4 | 0.618 ± 0.139 | 6.094 ± 0.226 | 4.247 ± 0.469 | 0.00% ± 0.00 |
| Rank32 × 0.75 | 4 | 0.782 ± 0.158 | 6.201 ± 0.127 | 4.434 ± 0.264 | 0.00% ± 0.00 |
| Rank32 × 1.0 | 4 | 0.776 ± 0.142 | 6.092 ± 0.131 | 4.114 ± 0.178 | 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 Affection. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Affection. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] La puerta sigue abierta. (Intense Affection. 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 Affection. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Dime cuándo estás listo. | Target 0.866 · CE 6.05 · Genuine 4.09 · WER 0.0% | Target 0.503 · CE 5.90 · Genuine 3.79 · WER 0.0% | Target 0.571 · CE 6.17 · Genuine 4.31 · WER 0.0% | Target 0.761 · CE 5.97 · Genuine 4.06 · 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 Affection. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Affection. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] La puerta sigue abierta. (Intense Affection. 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 Affection. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Dime cuándo estás listo. | Target 0.809 · CE 6.15 · Genuine 4.34 · WER 0.0% | Target 0.494 · CE 5.90 · Genuine 3.89 · WER 0.0% | Target 0.757 · CE 6.04 · Genuine 4.12 · WER 0.0% | Target 0.593 · CE 6.01 · Genuine 3.98 · 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 Affection. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Affection. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] He encontrado la cajita. (Intense Affection. 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 Affection. 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 0.542 · CE 6.12 · Genuine 4.70 · WER 0.0% | Target 0.758 · CE 6.32 · Genuine 4.65 · WER 0.0% | Target 0.873 · CE 6.28 · Genuine 4.62 · WER 0.0% | Target 0.934 · CE 6.14 · Genuine 4.38 · 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 Affection. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Affection. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] He encontrado la cajita. (Intense Affection. 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 Affection. 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 0.626 · CE 5.99 · Genuine 4.41 · WER 0.0% | Target 0.716 · CE 6.26 · Genuine 4.66 · WER 0.0% | Target 0.927 · CE 6.32 · Genuine 4.68 · WER 0.0% | Target 0.816 · CE 6.25 · Genuine 4.04 · WER 0.0% |