Child Automated Speech-to-Text Team (CAST)

The focus of this project is developing an automated speech-to-text program that works with young children (ages 3-5). The team will develop the algorithms for transcription and a functional interface that is user-friendly.



Many social science and education fields engage in significant research on parent-child verbal interactions. Unfortunately, many barriers exist to efficiently completing this research such as the cumbersome nature of transcribing (typically by hand) the verbal interactions that are observed or recorded during such research. Although much work has been done in adult speech-to-text, far less has been conducted with young children because of the rapidly changing nature of their speech. Using existing speech datasets, this team will develop an automated speech-to-text algorithm that works with young children (3-5 years old) and create a functional user interface.

During the 2022-2023 academic year, the CAST team will focus on iterating on the machine learning models that have been developed thus far as well as developing new algorithms that will help complete the project. The team will also begin development of a front-end user interface that will house the speech-to-text model.

Video Links:

Relevant Technologies:

  • Machine Learning
  • Natural Language Processing
  • Signal Processing


Python or other related programming language

Meeting time:

  • Spring 2022: Mondays, Wednesdays & Fridays 10:30-11:20 am, EE 011
  • Fall 2022: Mondays, Wednesdays & Fridays 10:30-11:20 am, EE 013