VMEF AI - Interactive 3D AI Companions & Knowledge Center

Welcome to VMEF AI, the full-stack ecosystem for persistent companion memory, real-time 3D VRM avatars, neural voice synthesis, and decentralized character creation.

1. Executive Platform Overview

VMEF AI is a next-generation web platform engineered to deliver deep, immersive, and persistent digital relationships with artificial intelligence. While typical chatbot applications operate statelessly—forgetting user names, personal preferences, and shared milestones as soon as a session concludes or exceeds context limits—VMEF AI establishes continuous memory streams that evolve dynamically over time.

Our platform unites four core technological breakthroughs:

  • The Soul Memory Engine: High-dimensional vector embedding extraction and semantic retrieval that keeps track of past interactions, hobbies, and emotional trajectories.
  • Multi-Provider Intelligence Backends: Seamless routing to leading language models including Anthropic Claude, OpenAI GPT-4, Google Gemini, and DeepSeek.
  • 3D glTF/VRM Avatar Rendering: WebGL 3D humanoid character rendering with procedural eye blinking, gaze tracking, spring bone hair physics, and live lip-syncing.
  • Neural Voice Cloning & Duplex Audio: Low-latency text-to-speech voice generation with ElevenLabs, Cartesia, and hands-free mid-sentence interruption.

2. Getting Started with VMEF AI

Setting up an account on VMEF AI takes less than two minutes. New users can authenticate using email or instant Google OAuth. Upon registration, your account is immediately credited with complimentary welcome Gold and Silver coins, allowing you to explore companion chats, test custom voices, and try 3D avatar viewport rendering right away.

Main Workspace Workflows:

  1. Discover (/discover): Browse community-created 3D AI characters categorized by genre, popularity, and creator badges.
  2. Chat Viewport (/chat): Interactive 3D avatar environment with live audio playback, memory inspections, and persona roleplay controls.
  3. Character Creator (/character-creator): A 5-step studio for designing unique AI personas, backstories, greetings, and visual styles.
  4. Voice Creator (/voice): Record directly through your microphone or upload audio files to clone custom neural voice profiles.
  5. Marketplace (/marketplace): Publish your original characters and voice models to earn coin royalties from community unlocks.

3. Soul Engine & Context Persistence Architecture

Standard stateless chatbots suffer from rapid memory decay. VMEF AI solves this fundamental limitation through a multi-tiered memory pipeline:

  • Working Memory: Immediate conversation context containing the most recent conversational exchanges and conversational momentum.
  • Episodic Long-Term Memory: Crucial facts, user hobbies, shared life stories, and personal preferences extracted and indexed into high-dimensional vector embeddings.
  • Semantic Context Injection: Before generating an AI response, relevant episodic memory nodes are retrieved via semantic similarity and injected into the character's system prompt.
  • Dynamic Emotional Posture: Emotional sentiment tracking across chat threads that allows companions to reflect appropriate emotional warmth, excitement, or empathy.

4. 3D VRM Avatar Rendering & Lip-Sync Physics

VMEF AI natively supports the open 3D VRM humanoid avatar standard (glTF 2.0 extension). Avatars render in real-time inside the browser using Three.js WebGL rendering without requiring desktop software or game engine plugins:

  • Real-Time Lip-Sync: Synthesized speech audio is analyzed live using the Web Audio API AnalyserNode. Viseme mouth shapes (aa, ih, ou, ee, oh) align directly with vocal output.
  • Spring Bone Physics: Secondary hair and clothing bone collision physics create realistic dynamic movement during idle animations.
  • Procedural Micro-Expressions: Natural blinking, subtle breathing cycles, and cursor-following eye gaze tracking for lifelike presence.
  • Model Creation Compatibility: Compatible with 3D avatars created in VRoid Studio, Blender VRM exporter, or custom rigged humanoid glTF assets.

5. Neural Voice Cloning & Duplex Speech Streaming

VMEF AI offers state-of-the-art voice generation and hands-free voice interaction:

  • Multi-Provider TTS Backends: Integration with ElevenLabs, Cartesia Sonic, Inworld AI, Fish Audio, and Google Cloud Speech.
  • Voice Sample Cloning: Upload a 30 to 60-second clear MP3/WAV recording to clone custom vocal identities with human-like warmth and pitch modulation.
  • Duplex Hands-Free Mode: Speak naturally through your microphone without holding buttons. If you speak while your companion is replying, speech playback halts instantly for natural mid-sentence interruption.
  • Speech-to-Text (STT) Recognition: Low-latency real-time voice transcription powered by AssemblyAI, ElevenLabs Scribe, and Deepgram.

6. 5-Step AI Character Creator Studio

The Character Creator Studio allows users to construct rich, original AI companions from scratch:

  1. Step 1: Visual Identity & Metadata: Define character name, category, avatar image, and background banners.
  2. Step 2: Persona & System Prompt: Write backstories, personality guidelines, negative constraints, and the initial greeting message.
  3. Step 3: Voice Integration: Assign pre-built neural voice profiles or link custom cloned voices with pitch/speed adjustments.
  4. Step 4: UI Themes & Viewport Rigging: Choose custom chat bubble themes, accent colors, and 3D avatar viewport lighting.
  5. Step 5: Publishing & Monetization Controls: Set visibility to Public SFW, Unlisted, or Private, and configure coin unlock pricing.

7. Virtual Economy & Marketplace Rewards

VMEF AI operates a dual-currency credit ecosystem designed to reward active community members and creators:

  • Silver Coins: Standard utility currency earned through consecutive daily login streaks, referral invitations, and community engagement.
  • Gold Coins: Premium currency used for custom voice cloning, high-tier 3D models, and premium AI backends.
  • Creator Royalties: Verified creators earn coin royalties whenever community members unlock or chat with their published creations.

8. Privacy, Data Protection & Publisher Compliance

We are committed to user privacy, data security, and publisher standards:

  • Encrypted Private Sessions: Private companion prompts, chat logs, and uploaded files are protected with TLS 1.3 encryption and are never exposed publicly or used for model training.
  • AES-256 Key Vault: User-provided API keys (OpenRouter, ElevenLabs) are encrypted at rest and used solely for your requests.
  • Automated SFW Moderation: Public marketplace creations undergo strict content moderation to ensure compliance with publisher advertising policies.

9. System Requirements & Support Contacts

VMEF AI runs smoothly across desktop and mobile browsers supporting WebGL 2.0 (Google Chrome 90+, Mozilla Firefox 88+, Apple Safari 14+, Microsoft Edge 90+).

For technical support, billing inquiries, or partnership opportunities: