Spaces:
Paused
Paused
| title: AGI Telecom POC | |
| emoji: 📡 | |
| colorFrom: blue | |
| colorTo: indigo | |
| sdk: docker | |
| sdk_version: "latest" | |
| app_file: app.py | |
| pinned: false | |
| # AGI Telecom POC | |
| This Hugging Face Space demonstrates an AGI-powered telecom interface that enables voice and text interaction through telecommunication channels (WebRTC/SIP). | |
| ## Overview | |
| This proof-of-concept showcases: | |
| - Multimodal communication (voice + text) | |
| - Agentic intelligence (reasoning, memory, response) | |
| - Telecom-enabled delivery (SIP/WebRTC) | |
| The system is powered by: | |
| - Meta-Llama-3.1-8B-Instruct through Hugging Face Inference Endpoints | |
| - Whisper for speech-to-text conversion | |
| - Edge TTS for natural-sounding speech synthesis | |
| ## Using the Interface | |
| This demo provides two ways to interact with the system: | |
| 1. **Web Interface**: A user-friendly chat interface with voice capabilities | |
| - Type messages or use voice input | |
| - See real-time visualizations of audio | |
| - Experience AI responses via text and speech | |
| 2. **API Endpoints**: Direct access for integration | |
| - `/query` - Process text with agent | |
| - `/transcribe` - Convert audio to text | |
| - `/speak` - Convert text to speech | |
| - `/complete_flow` - End-to-end processing | |
| ## Architecture | |
| The system follows this processing flow: |