Conversational AI Voice Agent for Mishkat Metaverse
The Future of Immersive Learning Begins with Natural Conversations
Antier engineered a bilingual conversational AI Voice Agent for Mishkat Metaverse, an immersive education platform commissioned by King Abdullah City for Atomic and Renewable Energy, supporting AI-powered learning at a national scale.
Business Challenge
Creating an intelligent guide for an immersive learning environment required far more than introducing conversational AI into a virtual platform. Mishkat Metaverse needed a voice-driven experience that felt natural, culturally relevant, and responsive. Achieving this vision required addressing four fundamental challenges.
Arabic First Voice Interaction
Support voice queries, AI responses, and subtitle overlays in modern standard Arabic as the primary language while providing English as a seamless fallback.
Three-Dimensional AI Persona
Move beyond a conventional chatbot by creating a culturally appropriate humanoid avatar with idle animations, natural lip synchronization, and a seamless presence within the virtual environment.
Sub Five Second End-to-End Latency
Deliver the complete Speech to Text, Large Language Model, and Text to Speech workflow in under five seconds to create natural and uninterrupted conversations.
Vendor Independent AI Architecture
Build an abstraction layer that enables AI model replacement through configuration without requiring client redeployment.
Business Objectives
Mishkat Metaverse sought to establish a voice-first learning environment that supports natural interactions. The initiative focused on creating an experience that balances accessibility, responsiveness, and operational flexibility.
Language Accessibility
Deliver a bilingual voice experience with Modern Standard Arabic as the primary language and English as a seamless secondary option.
Immersive User Engagement
Create a believable AI guide that becomes a natural part of the virtual learning environment.
Conversational Responsiveness
Maintain a fluid voice experience that keeps learners engaged throughout every interaction.
Operational Flexibility
Support future language model adoption through configuration without requiring changes to the client application.
Antier's Solution
Antier designed and developed a modular Conversational AI Voice Agent that operates natively within Mishkat Metaverse, enabling natural bilingual voice interactions across an immersive educational environment. Built on a server-side architecture, the solution delivers a scalable and maintainable foundation that supports long-term platform evolution while keeping the Unity WebGL client lightweight.
Solution Architecture
Antier engineered a four-stage AI pipeline that separates every stage of the conversational workflow into an independent component. This modular architecture keeps the Unity WebGL client lightweight while executing all intelligence on the server side.
Voice Capture
Learners initiate interactions through an in-platform microphone with a real-time waveform visualizer. A text input option automatically appears if voice recognition is unavailable.
Speech Recognition
Voice input is converted into text in Arabic or English through cloud-based speech recognition. The interaction language is selected according to the user preference configured during signup.
Conversational Intelligence
The language model processes each request and generates responses in the corresponding language through a vendor-independent abstraction layer that supports future model replacement through configuration.
Voice Response and Avatar Experience
Generated responses are converted into speech while the three-dimensional avatar performs synchronized lip movements. Dual language subtitles appear alongside the spoken response to create a natural conversational experience.
Key Capabilities
The Conversational AI Voice Agent was designed to deliver an intuitive and scalable learning experience while supporting the evolving requirements of the Mishkat Metaverse platform.
Arabic First Conversational Experience
Support natural voice interactions in Modern Standard Arabic with English available as a secondary language throughout the learning experience.
Embodied Three-Dimensional AI Guide
Deliver conversations through a culturally appropriate humanoid avatar with synchronized lip movements and natural idle animations.
Real Time Conversational Experience
Deliver fluid voice interactions through synchronized speech, avatar responses, and subtitle overlays that feel natural within the immersive environment.
Platform Connectivity
Operate alongside the platform analytics and gamification engines to support learner engagement and interaction tracking.
Performance Benchmarks
End-to-End AI Voice Response
Client Redeployments for Language Model Replacement
Maximum Subtitle Offset from Spoken Audio
Analytics Event Ingestion After Every Interaction
Business Outcomes
The Conversational AI Voice Agent established a scalable voice-first AI guide system that operates natively in Arabic while supporting long term platform growth.
Arabic Native Experience
Deliver Modern Standard Arabic voice responses, RTL subtitle overlays, and an Arabic first user experience from day one.
Operational Continuity
Enable underlying language model replacement through CMS configuration without engineer involvement, client redeployment, or platform downtime.
Complete Interaction Visibility
Stream every AI interaction anonymously to the analytics platform, enabling measurement of learner engagement, query patterns, and response quality from the beginning.
Project Highlights
Start a conversation with Antier
Connect with our consulting and engineering leads to scope your digital transformation from architecture review to production deployment.