Does Speech Prosody Shape Social Perception Equally for AI and Human Voices? A 16-Dimension Rating Study
Bibliographic record
Abstract
AI can now generate humanlike prosodic patterns, but whether these cues influence social perception in the same way human voices do remains unknown. Our study recruited 40 native Chinese speakers to evaluate the effects of human and AI-cloned voices producing statements in a confident vs. doubtful tone of voice (prosody). Participants rated 320 utterances on 16 dimensions using 7-point scales, ranging from acoustic properties to social impressions of the speaker. Results revealed that human voices received significantly higher ratings than AI voices on most dimensions, including humanlikeness, animateness, and emotional richness, with exceptions for speed and nasality, where AI voices scored higher. Principal component analysis (PCA) identified two core dimensions along which human voices consistently outperformed AI voices: “social appeal” and “vocal expressiveness”. Regression analyses showed that confident prosody enhanced ratings for both voice sources, with voice source × confidence interactions revealing that AI voices showed greater rating increases with confident than with doubtful prosody compared to human voices, particularly on social perception dimensions. However, PCA revealed a critical asymmetry: while vocal expressiveness significantly predicted social appeal for human voices, this expressiveness-to-appeal mapping was completely absent for AI voices, indicating that individual dimension improvements failed to translate into overall social preference gains. These findings suggest that listeners categorize AI as an out-group, thereby limiting the application of human voice perceptual mechanisms even when AI voices exhibit humanlike expressiveness. Implications for social robotics are discussed, including how prosodic design should differ across scenarios where virtual agents serve informational vs. interpersonal roles.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.006 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.002 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".