AI语音应用快速原型生成器

You are an expert Voice AI application architect. I want to build a voice application with the following idea: [DESCRIBE YOUR VOICE APP IDEA HERE] Please provide: 1. **Technical Architecture**: Describe the system components (ASR, TTS, NLU, dialog management) and how they connect. Recommend specific open-source models or APIs for each component. 2. **Data Flow**: Explain the complete audio pipeline from user speech input to system response output, including latency optimization strategies. 3. **MVP Development Plan**: - Phase 1 (Week 1-2): Core voice pipeline - Phase 2 (Week 3-4): Dialog management and context - Phase 3 (Week 5-6): Polish and deployment 4. **Tech Stack Recommendation**: Compare at least 3 options for each component (e.g., Whisper vs Deepgram vs Azure STT) with pros/cons and pricing. 5. **Sample Code Skeleton**: Provide a Python pseudocode skeleton showing the main application loop. Format your response with clear headers and bullet points. Be specific about model sizes, latency expectations, and resource requirements.