Voice Cloning Using AI vs Traditional Audio Recording for Prerecorded Courses in Medical Pedagogy: Randomized Controlled Trial
Antoine Gavoille e outros, 2026 · JMIR Medical Education
AI voice cloning in prerecorded medical lectures led to virtually identical knowledge test scores compared to human voice recordings.
Estimativa · total score on two online knowledge tests (acquisition and application), points out of 100
-0,900 [-2,70; 4,40]
points out of 100
P=.60
- Desenho
- Experimento aleatório. randomized controlled trial with 1:1 allocation stratified by study year
- População
- Fourth- and fifth-year medical students at the University of Lyon, France
- Amostra
- 50 · medical students in the mITT population who completed at least one knowledge test
Onde está o número
Results > Participants’ Results on the Knowledge Test Scores; section 'Participants’ Results on the Knowledge Test Scores'
The mean total score on the 2 knowledge tests was 51.2 (SD 4.9) points in the AI voice cloning group and 51.8 (SD 6.9) in the audio recording group, with no significant difference between the 2 groups (adjusted mean difference −0.9, 95% CI −2.7 to 4.4; P=.60).
Contexto
What changed: Four 10-minute online lectures used an AI-generated clone of the instructor's voice instead of conventional audio recordings. Comparison: Prerecorded lectures narrated by the teacher using conventional recording equipment. Limits: Single-institution trial with a modest sample size and missing test data for 22% of the modified intention-to-treat group.
Caminho no grafo
- FamíliaProvisão direta
- TipoProvisão de serviço público
- ResultadoAprendizado (proficiência em testes)
- EixoEducação
MAGI só cita evidência deste instrumento
Lugares do estudo
- France · País · FR
- Fonte
- DOI: 10.2196/86569 Fonte
- Licença
- CC-BY
- Revisão
- Extraído por máquina, não revisadoExtraído por: gemini-3.8-flash, 2026-10-23