TeacherPrints

UC Berkeley | Capstone
Futures Forum on Learning | Tools Competition Catalyst Prize Winner
Anna Jacobson, Tim Slade, Lee Moore, and Kim Darnell

---

Project Description

An end-to-end data science product designed to empower teachers around the world to envision teaching excellence through accessible, affordable instructional coaching, available anytime and anywhere.

Skills

Machine Learning (Automatic Speech Recognition), Data Analysis, Data Visualization, Web Design, Written Communication, Professional Presentation, Research, User Testing

Tools

Python (Pyannote-Audio/Pytorch), D3.js, HTML/CSS (Bootstrap), Flask, AWS (EC2, S3, and RDS), Jupyter Notebook, Github, Photoshop, Google Slides

---

Motivation

TeacherPrints is a free instructional coaching app that utilizes personalized, data-driven visualizations to help teachers bring the reality of their lesson delivery in line with their expectations and goals. We apply Pytorch-based voice-type classification and speech activity detection deep learning models to audio recordings of instructional interaction in the classroom. Through a combination of Python for analysis and d3.js for visualization, we transform these data into an intuitive, customized summary of educationally important, talk-related features. Guided professional reflection and curated online resources help teachers translate TeacherPrints insights into actionable next steps for their class. Regular use of TeacherPrints allows teachers to document their progress and identify new instructional goals along their path to teaching excellence.

---