HeyGen ships GPT-Live language tutor: full-duplex speech model drives an avatar teacher
HeyGen · x · 2026-10-06
HeyGen released a reference implementation connecting OpenAI GPT-Live, a full-duplex speech-to-speech model, to its LiveAvatar, rendering the model's tool calls as animated overlays on the avatar's video. Out of the box it's a Japanese tutor that teaches vocabulary aloud with on-screen term cards (word · pronunciation · meaning) and periodically shrinks to a corner to review. A Node orchestrator bridges GPT-Live's audio stream into the avatar's media server; personas are just two Markdown files. The repo is MIT-licensed. The post cites Bloom's 1984 finding that 1:1 tutoring outperformed 98% of a conventional class, arguing AI finally makes tutors universal for language learners.
More from coding & agent
- Dev Experiments With an Artifacts Tab: Let Agents Package Work as Visual Overviews — charlieholtz · 2026-10-06
- User burns $107 in 90 minutes on Claude Code Cloud, billed like an API key — MicahBerkley · 2026-10-06
- Google Docs/Drive beats local markdown for keeping agent context in sync, dev finds — AI_Andrew · 2026-10-06
- Anthropic shifts Cowork task execution to the cloud, engineer explains why — lydiahallie · 2026-10-06
- Cursor SDK subagents now report back to parent, system prompt becomes replaceable — tetsuoai · 2026-10-06
- Grok Bot announces six live workshops in two weeks, covering marketing, sales and coding — XFreeze · 2026-10-06