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 | -0.105 | 0.263 | 6.261 | 4.357 | 0.597 | 0.011 | 1.000 | pass |
| manual | 50 | 1.052 | 0.822 | 5.916 | 4.541 | 0.541 | 0.032 | 0.940 | fail |
| adapter | 50 | 1.191 | 0.890 | 5.902 | 4.538 | 0.535 | 0.027 | 0.940 | fail |
| matched_no_adapter | 50 | 1.052 | 0.822 | 5.916 | 4.541 | 0.541 | 0.032 | 0.940 | fail |
Incremental adapter gain versus same-CFG control: 0.139 z;95% cluster interval [-0.10713867753872283, 0.3856347573166811]. Versus best manual-only: 0.139 z. Manual-only versus standard: 1.157 z.
| Scale | CFG | Target z | CE | Genuine | WER | Provisional QA |
|---|---|---|---|---|---|---|
| 0.0 | 1.0 | 0.141 | 6.120 | 4.432 | 0.005 | True |
| 0.5 | 1.0 | 0.180 | 6.140 | 4.464 | 0.009 | True |
| 0.75 | 1.0 | 0.314 | 6.141 | 4.466 | 0.003 | True |
| 1.0 | 1.0 | 0.021 | 6.197 | 4.496 | 0.005 | True |
| 0.0 | 1.25 | 0.342 | 6.156 | 4.520 | 0.011 | True |
| 0.5 | 1.25 | 0.356 | 6.155 | 4.469 | 0.003 | True |
| 0.75 | 1.25 | 0.300 | 6.135 | 4.498 | 0.003 | True |
| 1.0 | 1.25 | 0.414 | 6.123 | 4.462 | 0.000 | True |
| 0.0 | 1.5 | 0.528 | 6.118 | 4.486 | 0.000 | True |
| 0.5 | 1.5 | 0.323 | 6.138 | 4.467 | 0.003 | True |
| 0.75 | 1.5 | 0.473 | 6.133 | 4.553 | 0.005 | True |
| 1.0 | 1.5 | 0.565 | 6.125 | 4.553 | 0.000 | True |
| 0.0 | 2.0 | 0.816 | 6.048 | 4.509 | 0.019 | True |
| 0.5 | 2.0 | 0.658 | 6.077 | 4.559 | 0.022 | True |
| 0.75 | 2.0 | 0.586 | 6.076 | 4.541 | 0.000 | True |
| 1.0 | 2.0 | 0.569 | 6.032 | 4.478 | 0.008 | True |
| 0.0 | 2.5 | 0.935 | 6.027 | 4.563 | 0.009 | True |
| 0.5 | 2.5 | 0.808 | 6.067 | 4.593 | 0.006 | True |
| 0.75 | 2.5 | 0.771 | 6.004 | 4.617 | 0.003 | True |
| 1.0 | 2.5 | 1.058 | 6.012 | 4.568 | 0.017 | True |
| 0.0 | 3.0 | 1.131 | 5.977 | 4.595 | 0.011 | True |
| 0.5 | 3.0 | 0.941 | 6.024 | 4.570 | 0.015 | True |
| 0.75 | 3.0 | 1.172 | 5.972 | 4.646 | 0.038 | True |
| 1.0 | 3.0 | 0.966 | 6.023 | 4.591 | 0.008 | True |
| 0.0 | 4.0 | 1.225 | 5.821 | 4.584 | 0.008 | True |
| 0.5 | 4.0 | 1.271 | 5.864 | 4.598 | 0.032 | True |
| 0.75 | 4.0 | 1.292 | 5.839 | 4.572 | 0.025 | True |
| 1.0 | 4.0 | 1.281 | 5.932 | 4.609 | 0.006 | True |
| 0.0 | 5.0 | 1.281 | 5.809 | 4.577 | 0.031 | True |
| 0.5 | 5.0 | 1.206 | 5.810 | 4.567 | 0.011 | False |
| 0.75 | 5.0 | 1.317 | 5.776 | 4.562 | 0.014 | True |
| 1.0 | 5.0 | 1.252 | 5.751 | 4.503 | 0.008 | False |
direct_manual_recipe
{
"id": "manual-93bbe240eddf",
"general": "Focus on vocal texture to create a sense of fragile hesitation, as if the voice itself is wavering.",
"inline": [
"Maintain a slightly breathy quality throughout the first phrase, implying lack of air support.",
"Introduce a momentary vocal fry or slight catch in the second segment for emphasis.",
"Deliver the final phrase with a thin, slightly brittle resonance."
],
"cfg": 3,
"recipe": "P07",
"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 Doubt. 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": 8, "de": 6, "es": 6, "fr": 6, "nl": 4}. 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 | 0.440 ± 0.222 | 6.201 ± 0.146 | 4.383 ± 0.244 | 0.00% ± 0.00 |
| Rank32 × 0.5 | 20 | 0.472 ± 0.243 | 6.179 ± 0.133 | 4.383 ± 0.239 | 0.00% ± 0.00 |
| Rank32 × 0.75 | 20 | 0.467 ± 0.255 | 6.160 ± 0.149 | 4.330 ± 0.131 | 0.28% ± 1.24 |
| Rank32 × 1.0 | 20 | 0.434 ± 0.226 | 6.200 ± 0.144 | 4.366 ± 0.220 | 0.00% ± 0.00 |
Fixed speaker reference:
| Condition | n | Target score ↑ | Content Enjoyment ↑ | Genuineness ↑ | WER ↓ |
|---|---|---|---|---|---|
| Merged base, no bucket adapter | 4 | 0.324 ± 0.043 | 6.283 ± 0.123 | 4.179 ± 0.316 | 0.00% ± 0.00 |
| Rank32 × 0.5 | 4 | 0.324 ± 0.042 | 6.305 ± 0.094 | 4.201 ± 0.252 | 0.00% ± 0.00 |
| Rank32 × 0.75 | 4 | 0.376 ± 0.118 | 6.249 ± 0.103 | 4.280 ± 0.223 | 0.00% ± 0.00 |
| Rank32 × 1.0 | 4 | 0.339 ± 0.056 | 6.301 ± 0.110 | 4.289 ± 0.303 | 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 Doubt. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Doubt. 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 Doubt. 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 Doubt. 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 0.292 · CE 6.21 · Genuine 4.01 · WER 0.0% | Target 0.290 · CE 6.21 · Genuine 3.92 · WER 0.0% | Target 0.288 · CE 6.22 · Genuine 4.03 · WER 0.0% | Target 0.300 · CE 6.21 · Genuine 4.03 · 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 Doubt. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Doubt. 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 Doubt. 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 Doubt. 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 0.296 · CE 6.17 · Genuine 3.84 · WER 0.0% | Target 0.287 · CE 6.24 · Genuine 4.06 · WER 0.0% | Target 0.280 · CE 6.14 · Genuine 4.18 · WER 0.0% | Target 0.286 · CE 6.20 · Genuine 4.02 · 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 Doubt. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Doubt. 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 Doubt. 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 Doubt. 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 0.385 · CE 6.31 · Genuine 4.55 · WER 0.0% | Target 0.347 · CE 6.36 · Genuine 4.42 · WER 0.0% | Target 0.404 · CE 6.39 · Genuine 4.55 · WER 0.0% | Target 0.405 · CE 6.40 · Genuine 4.57 · 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 Doubt. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Doubt. 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 Doubt. 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 Doubt. 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 0.321 · CE 6.44 · Genuine 4.33 · WER 0.0% | Target 0.371 · CE 6.40 · Genuine 4.41 · WER 0.0% | Target 0.532 · CE 6.25 · Genuine 4.36 · WER 0.0% | Target 0.365 · CE 6.39 · Genuine 4.53 · WER 0.0% |
Fixed speaker reference:
| Condition | n | Target score ↑ | Content Enjoyment ↑ | Genuineness ↑ | WER ↓ |
|---|---|---|---|---|---|
| Merged base, no bucket adapter | 4 | 0.319 ± 0.025 | 6.372 ± 0.069 | 4.504 ± 0.043 | 0.00% ± 0.00 |
| Rank32 × 0.5 | 4 | 0.420 ± 0.166 | 6.297 ± 0.134 | 4.357 ± 0.128 | 0.00% ± 0.00 |
| Rank32 × 0.75 | 4 | 0.302 ± 0.043 | 6.293 ± 0.154 | 4.297 ± 0.043 | 0.00% ± 0.00 |
| Rank32 × 1.0 | 4 | 0.409 ± 0.152 | 6.358 ± 0.112 | 4.454 ± 0.220 | 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 Doubt. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Doubt. 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 Doubt. 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 Doubt. 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 0.325 · CE 6.28 · Genuine 4.47 · WER 0.0% | Target 0.637 · CE 6.27 · Genuine 4.36 · WER 0.0% | Target 0.365 · CE 6.11 · Genuine 4.36 · WER 0.0% | Target 0.315 · CE 6.26 · Genuine 4.32 · 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 Doubt. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Doubt. 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 Doubt. 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 Doubt. 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 0.342 · CE 6.35 · Genuine 4.52 · WER 0.0% | Target 0.444 · CE 6.13 · Genuine 4.48 · WER 0.0% | Target 0.275 · CE 6.22 · Genuine 4.26 · WER 0.0% | Target 0.636 · CE 6.38 · Genuine 4.78 · 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 Doubt. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Doubt. 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 Doubt. 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 Doubt. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Wir können gemeinsam hineinschauen. | Target 0.284 · CE 6.43 · Genuine 4.56 · WER 0.0% | Target 0.350 · CE 6.35 · Genuine 4.41 · WER 0.0% | Target 0.271 · CE 6.44 · Genuine 4.29 · WER 0.0% | Target 0.361 · CE 6.51 · Genuine 4.36 · 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 Doubt. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Doubt. 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 Doubt. 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 Doubt. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Wir können gemeinsam hineinschauen. | Target 0.327 · CE 6.43 · Genuine 4.47 · WER 0.0% | Target 0.248 · CE 6.44 · Genuine 4.18 · WER 0.0% | Target 0.296 · CE 6.40 · Genuine 4.28 · WER 0.0% | Target 0.325 · CE 6.28 · Genuine 4.36 · WER 0.0% |
Fixed speaker reference:
| Condition | n | Target score ↑ | Content Enjoyment ↑ | Genuineness ↑ | WER ↓ |
|---|---|---|---|---|---|
| Merged base, no bucket adapter | 4 | 0.404 ± 0.117 | 6.094 ± 0.140 | 4.498 ± 0.171 | 0.00% ± 0.00 |
| Rank32 × 0.5 | 4 | 0.387 ± 0.096 | 6.114 ± 0.083 | 4.589 ± 0.081 | 0.00% ± 0.00 |
| Rank32 × 0.75 | 4 | 0.392 ± 0.135 | 6.179 ± 0.115 | 4.412 ± 0.083 | 0.00% ± 0.00 |
| Rank32 × 1.0 | 4 | 0.295 ± 0.014 | 6.176 ± 0.068 | 4.374 ± 0.155 | 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 Doubt. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Doubt. 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 Doubt. 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 Doubt. 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 0.408 · CE 6.05 · Genuine 4.66 · WER 0.0% | Target 0.372 · CE 6.10 · Genuine 4.51 · WER 0.0% | Target 0.320 · CE 6.16 · Genuine 4.43 · WER 0.0% | Target 0.290 · CE 6.23 · Genuine 4.56 · 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 Doubt. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Doubt. 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 Doubt. 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 Doubt. 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 0.326 · CE 6.13 · Genuine 4.33 · WER 0.0% | Target 0.306 · CE 6.23 · Genuine 4.53 · WER 0.0% | Target 0.590 · CE 6.05 · Genuine 4.31 · WER 0.0% | Target 0.282 · CE 6.18 · Genuine 4.19 · 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 Doubt. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Doubt. 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 Doubt. 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 Doubt. 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 0.314 · CE 6.26 · Genuine 4.37 · WER 0.0% | Target 0.344 · CE 6.09 · Genuine 4.65 · WER 0.0% | Target 0.299 · CE 6.33 · Genuine 4.40 · WER 0.0% | Target 0.295 · CE 6.08 · Genuine 4.32 · 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 Doubt. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Doubt. 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 Doubt. 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 Doubt. 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 0.568 · CE 5.93 · Genuine 4.63 · WER 0.0% | Target 0.524 · CE 6.03 · Genuine 4.67 · WER 0.0% | Target 0.358 · CE 6.17 · Genuine 4.51 · WER 0.0% | Target 0.313 · CE 6.21 · Genuine 4.43 · WER 0.0% |
Fixed speaker reference:
| Condition | n | Target score ↑ | Content Enjoyment ↑ | Genuineness ↑ | WER ↓ |
|---|---|---|---|---|---|
| Merged base, no bucket adapter | 4 | 0.836 ± 0.168 | 6.122 ± 0.077 | 4.396 ± 0.128 | 0.00% ± 0.00 |
| Rank32 × 0.5 | 4 | 0.900 ± 0.153 | 6.088 ± 0.098 | 4.361 ± 0.068 | 0.00% ± 0.00 |
| Rank32 × 0.75 | 4 | 0.933 ± 0.089 | 6.106 ± 0.060 | 4.365 ± 0.075 | 1.39% ± 2.78 |
| Rank32 × 1.0 | 4 | 0.811 ± 0.221 | 6.126 ± 0.042 | 4.302 ± 0.071 | 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 Doubt. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Doubt. 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 Doubt. 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 Doubt. 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 0.981 · CE 6.07 · Genuine 4.52 · WER 0.0% | Target 0.994 · CE 6.15 · Genuine 4.27 · WER 0.0% | Target 0.974 · CE 6.13 · Genuine 4.46 · WER 0.0% | Target 1.000 · CE 6.13 · Genuine 4.31 · 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 Doubt. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Doubt. 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 Doubt. 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 Doubt. 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 0.982 · CE 6.06 · Genuine 4.40 · WER 0.0% | Target 0.992 · CE 5.94 · Genuine 4.38 · WER 0.0% | Target 0.989 · CE 6.03 · Genuine 4.31 · WER 0.0% | Target 0.996 · CE 6.18 · Genuine 4.36 · 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 Doubt. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Doubt. 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 Doubt. 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 Doubt. 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 0.689 · CE 6.13 · Genuine 4.22 · WER 0.0% | Target 0.939 · CE 6.11 · Genuine 4.43 · WER 0.0% | Target 0.800 · CE 6.17 · Genuine 4.39 · WER 5.6% | Target 0.564 · CE 6.09 · Genuine 4.20 · 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 Doubt. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Doubt. 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 Doubt. 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 Doubt. 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 0.691 · CE 6.23 · Genuine 4.44 · WER 0.0% | Target 0.674 · CE 6.15 · Genuine 4.36 · WER 0.0% | Target 0.969 · CE 6.09 · Genuine 4.30 · WER 0.0% | Target 0.685 · CE 6.10 · Genuine 4.34 · WER 0.0% |
Fixed speaker reference:
| Condition | n | Target score ↑ | Content Enjoyment ↑ | Genuineness ↑ | WER ↓ |
|---|---|---|---|---|---|
| Merged base, no bucket adapter | 4 | 0.317 ± 0.018 | 6.132 ± 0.107 | 4.338 ± 0.368 | 0.00% ± 0.00 |
| Rank32 × 0.5 | 4 | 0.332 ± 0.033 | 6.092 ± 0.040 | 4.407 ± 0.411 | 0.00% ± 0.00 |
| Rank32 × 0.75 | 4 | 0.332 ± 0.019 | 5.973 ± 0.065 | 4.297 ± 0.167 | 0.00% ± 0.00 |
| Rank32 × 1.0 | 4 | 0.317 ± 0.016 | 6.040 ± 0.104 | 4.408 ± 0.334 | 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 Doubt. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Doubt. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] La puerta sigue abierta. (Intense Doubt. 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 Doubt. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Dime cuándo estás listo. | Target 0.299 · CE 6.00 · Genuine 4.09 · WER 0.0% | Target 0.290 · CE 6.04 · Genuine 4.16 · WER 0.0% | Target 0.361 · CE 5.94 · Genuine 4.29 · WER 0.0% | Target 0.332 · CE 5.97 · Genuine 4.34 · 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 Doubt. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Doubt. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] La puerta sigue abierta. (Intense Doubt. 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 Doubt. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] Dime cuándo estás listo. | Target 0.303 · CE 6.13 · Genuine 3.96 · WER 0.0% | Target 0.344 · CE 6.12 · Genuine 3.96 · WER 0.0% | Target 0.321 · CE 5.94 · Genuine 4.07 · WER 0.0% | Target 0.298 · CE 5.94 · Genuine 3.97 · 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 Doubt. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Doubt. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] He encontrado la cajita. (Intense Doubt. 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 Doubt. 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 0.332 · CE 6.26 · Genuine 4.70 · WER 0.0% | Target 0.367 · CE 6.09 · Genuine 4.82 · WER 0.0% | Target 0.324 · CE 5.95 · Genuine 4.37 · WER 0.0% | Target 0.311 · CE 6.09 · Genuine 4.62 · 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 Doubt. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word. SCRIPT: (Intense Doubt. Make the requested delivery clearly audible, not merely implied by the words. Preserve every word.) [3.2 seconds duration] He encontrado la cajita. (Intense Doubt. 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 Doubt. 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 0.333 · CE 6.14 · Genuine 4.61 · WER 0.0% | Target 0.328 · CE 6.13 · Genuine 4.68 · WER 0.0% | Target 0.324 · CE 6.07 · Genuine 4.46 · WER 0.0% | Target 0.327 · CE 6.17 · Genuine 4.71 · WER 0.0% |