An automated pipeline for efficiently generating standardized, child-friendly audiovisual language stimuli

Area of Science:

• Cognitive Science
• Developmental Psychology
• Computational Linguistics

Background:

• Creating child-appropriate language stimuli is challenging and labor-intensive.
• Existing methods for stimulus generation can be slow and costly.

Purpose of the Study:

• To develop an automated pipeline for rapid, customizable audiovisual language stimulus generation.
• To simplify and accelerate the creation of stimuli for developmental and cognitive research.

Main Methods:

• An automated pipeline combining Google Cloud Text-to-Speech and Adobe Character Animator was developed.
• The pipeline converts text input into audiovisual stimuli with animated characters and synchronized speech.
• Stimuli were evaluated through acoustic comparison with human-recorded data.

Main Results:

• The pipeline efficiently generates audiovisual stimuli in under 2 minutes per item.
• The audio generation component is exceptionally fast, completing in under 1 second per stimulus.
• Less than 9% of generated stimuli required regeneration, indicating high quality.

Conclusions:

• The automated pipeline significantly streamlines the creation of language stimuli for research.
• This tool is particularly beneficial for cognitive neuroscience studies requiring large stimulus sets.
• The pipeline facilitates developmental research across language and other cognitive domains.