Download analysis/06_voice_study/05_replication.py from sciminds/same-words-different-judgements: direct link, hf CLI and curl.
- Browser
- Download file 2.15 kB
-
https://huggingface.co/datasets/sciminds/same-words-different-judgements/resolve/main/analysis/06_voice_study/05_replication.py
- Command line
-
hf download hf://datasets/sciminds/same-words-different-judgements/analysis/06_voice_study/05_replication.py
-
curl -L -o 05_replication.py https://huggingface.co/datasets/sciminds/same-words-different-judgements/resolve/main/analysis/06_voice_study/05_replication.py
2.15 kB
| """Do the paper's audio-side findings survive a change of voice? | |
| python analysis/06_voice_study/05_replication.py | |
| python analysis/06_voice_study/05_replication.py --skip-models # fast | |
| The voice study holds the words fixed and swaps the voice. The paper's claim is not | |
| "audio ratings are high" but "audio judgments are behaviourally distinct from | |
| text": raters commit to a winner at smaller gaps, lean on length less, favour | |
| the second-presented item, and agree with the text winner at near chance. If | |
| those signatures only exist for Kokoro af_heart they are properties of one TTS | |
| voice, not of the audio modality. This script recomputes each of them per arm. | |
| Five quantities, matching Section 4 of the paper: | |
| threshold Mean per-prompt rating gap on non-tie trials (paper Fig. 1): | |
| audio 26.0 vs text 39.8 over 100 prompts. The paper's threshold | |
| text and tables pool trials instead (audio 27.9, text 41.7); | |
| 08_cross_study/01_threshold_summary.py reports that version. | |
| length Coefficient on z-scored response character length (paper Table 1; | |
| audio b = 3.52, text b = 4.92). | |
| recency Coefficient on presented-first (paper b = -2.61: first-presented | |
| rated LOWER, i.e. the second item is favoured). | |
| icc ICC(2,1) and ICC(2,k) from crossed random effects | |
| (paper audio .333 / .821, text .295 / .788). | |
| agreement Winner agreement with the ORIGINAL TEXT condition, per prompt. | |
| Every arm-level number is also recomputed for the original af_heart audio and | |
| the original text on the SAME 30 prompts. The published values cover 100 | |
| prompts, so comparing a 30-prompt arm against them would confound the voice | |
| with the stimulus subset. The matched columns are the honest comparison; the | |
| published ones are shown only for orientation. | |
| """ | |
| # Implementation lives in _voice_variation_replication.py, which other analyses import. | |
| import sys | |
| from pathlib import Path | |
| sys.path.insert(0, str(Path(__file__).resolve().parents[1] / "common")) | |
| from _voice_variation_replication import main # noqa: E402 | |
| if __name__ == "__main__": | |
| main() | |