Curated Learning Path for Building Real-Time Voice AI Agents

A GitHub repository titled 'Voice-AI-for-Beginners' offers a structured, developer-friendly learning path for building real-time voice AI agents, from initial speech-to-text calls to scaling production telephony. The resource, curated by mahimairaja, is organized into 20 sections covering foundational concepts, frameworks, components (STT, TTS, LLM, VAD), transport (WebRTC, telephony), production, ethics, and community resources. It recommends starting with LiveKit Agents or Pipecat as open-source frameworks, and provides a suggested 5-week learning path. The guide emphasizes latency budgets, turn-taking, and evaluation metrics, and includes vendor-neutral comparisons. It also covers regulatory aspects like the FCC's 2024 ruling on AI-generated voices in robocalls and the EU AI Act. The repository is actively maintained and accepts pull requests for resources that are less than 12 months old.