v1.8
CurrentMeet Virtual Ashesh
The portfolio now includes a voice-driven guide that can answer questions, navigate the site, and respond through an expressive 3D avatar.
- Added
A site-wide Virtual Ashesh interface with browser speech recognition, natural male speech synthesis, and concise spoken answers grounded in portfolio data.
- Added
An expressive 3D avatar with cursor-following eyes and head movement, natural blinking, emoji expressions, and word-boundary mouth animation.
- Added
Voice commands that reuse the portfolio command system to navigate routes and perform supported actions without losing the active guide session.
- Added
A server-only Gemini guide endpoint with cached portfolio context, FAQ shortcuts, conversation history, same-origin enforcement, and per-IP rate limiting.
- Improved
Avatar rendering and its model remain outside the initial page load through interaction-triggered code splitting and immutable model caching.
- Improved
Silent sessions receive spoken nudges before automatically closing, while manual closing immediately stops speech, listening, timers, and pending work.