README.md

metadata

title: AmaniQuery VibeVoice
emoji: 🎤
colorFrom: purple
colorTo: pink
sdk: docker
app_port: 7860
pinned: false
license: mit

AmaniQuery VibeVoice

Text-to-Speech service for AmaniQuery using Microsoft VibeVoice model.

Features

• 🎤 High-quality text-to-speech synthesis
• 💬 Conversational voice responses
• 🔊 Multiple voice presets (Wayne, Angela, etc.)
• ⚡ Real-time streaming audio

API Endpoints

• GET /health - Health check
• POST /api/v1/voice/speak - Convert text to speech
• POST /api/v1/voice/chat - Conversational voice synthesis
• GET /api/v1/voice/voices - List available voice presets

Request Format

{
  "text": "Hello, I am your legal research assistant.",
  "voice": "Wayne",
  "cfg_scale": 1.5
}

Environment Variables

Required secrets in HF Space settings:

• JWT_SECRET - Shared JWT secret for cross-service auth
• BACKEND_SPACE_URL - URL to main AmaniQuery backend