# Eeva presenter asset

Generated with the built-in image generation tool. Original fictional female presenter; not an imitation of the reference person. The website adds the button, disclosure and interactive controls in HTML/CSS.

## Final generation prompt

Use case: photorealistic-natural. Asset type: website AI assistant presenter portrait. Create an original fictional adult woman, early 30s, warm medium Indian complexion, shoulder-length softly waved dark brown hair, approachable natural smile, looking directly at camera. Wearing a beautifully tailored black blazer over a plain white crew-neck shirt and small understated gold earrings. Professional yet friendly, natural skin texture, realistic studio portrait photography, soft flattering light with a very subtle lavender rim light. Vertical waist-up composition, entire head and both shoulders inside frame, arms relaxed below frame. Centered subject filling most of canvas, modest headroom. Genuinely transparent background with clean alpha edges around hair and clothing, no backdrop or ground shadow. This is a cutout asset for a floating website assistant against a dark background. No text, buttons, UI, logos, watermark, props, borders, or other people. Do not imitate any specific real person's identity.

## Integration

Asset: public_html/eeva/media/eeva-presenter.png

The presenter is a still portrait. Speaking/listening wave indicators are animated; no facial animation or lip-synced video is included. The existing microphone and audio playback power voice conversations.
Voice update: replies automatically request speech unless muted. The browser audio context is unlocked on interaction; Replay remains available when autoplay is blocked. Listening and speaking bars follow microphone/playback levels, with a separate thinking animation. Voice tests use synthetic audio and mocked provider responses. Real provider audio still requires the configured OpenAI key.