-
Theta EdgeCloud:
- LLM/GenAI: We utilize a Large Language Model (LLM) to build a Retrieval-Augmented Generation (RAG) system. This system effectively leverages product and brand information to pitch products and answer viewer queries.
- Cloud Rendering: To provide high-quality, aesthetically pleasing video ads, we perform real-time 3D rendering of the ads on Theta EdgeCloud’s GPU-based instances, ensuring cost-effectiveness and scalability.
- STT/TTS: Speech-to-Text (STT) and Text-to-Speech (TTS) models are employed to generate and deliver responses in real-time.
- Live Stream: By utilizing Theta Edge nodes distributed globally, we achieve low latency and cost-effective live streaming of the 3D rendered video to the audience.
-
Theta Blockchain:
- We have deployed a smart contract on the Theta testnet to manage advertising campaigns and billing. This ensures transparency, convenience, and reliability in billing processes.
-
Unity3D:
- The 3D environment for the ads is developed using Unity3D. The environment includes various interactive elements and logic, exported as an executable. This executable runs on EdgeCloud for real-time rendering based on inputs from the core server.
-
API/Core:
- Serving as the brain of the system, our API/Core coordinates the various components. It determines which ads to show, manages interactions with the RAG system, handles cloud rendering, and ensures seamless integration of STT/TTS services. This ensures optimal performance and minimal latency.
ssd39/humaads.live
Folders and files
| Name | Name | Last commit date | ||
|---|---|---|---|---|
