We need to talk about accuracy and bias in voice tech for kids
summary
Company builds a child‑first speech engine that powers both play and education, licensed into products for classrooms and toys. Core claim is an independently validated accuracy profile with no bias in scoring across Black White Latino and high versus low poverty students. Engine runs on cheap phones tablets and toys in noisy rooms, provides voice activity detection and custom wake words, and performs fluency assessment that detects repetitions insertions substitutions deletions and returns per‑student reading scores.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 1:36Why voice tech for kids
- 2:26Differences between kids and adults
- 5:00Accuracy
- 11:21Bias Mitigation
- 12:22Voice Engine
- 13:55COVID
related talks
50mEthical Machine Learning
43mTesting Agentic AI Applications: Beyond Traditional QA
46mIntroduction to Speech Recognition Systems Through a Tester’s Eyes
46mTalking the Talk: Measuring Chatbot Accuracy
47mTesting Conversational AI
56mAutomating Voice Assistants and Chatbots
06m“Testing applications for children”
19mTesting LLM-Powered Applications