voice-realtime

Real-time voice AI knowledge including STT and TTS providers, LiveKit Agents plugins, and voice pipeline patterns. Use when working with speech-to-text, text-to-speech, voice agents, LiveKit, or any voice-related infrastructure.

⭐0stars🔀0forks📁View on GitHub🕐Updated Dec 26, 2025⬇️2downloads

When & Why to Use This Skill

This Claude skill provides comprehensive technical expertise for building and optimizing real-time voice AI systems. It offers deep insights into Speech-to-Text (STT) and Text-to-Speech (TTS) provider selection, LiveKit Agents integration, and voice pipeline architecture. By leveraging a detailed decision matrix, it helps developers balance latency, accuracy, and cost to create high-performance voice-enabled applications.

Use Cases

  • Architecting Voice Agents: Designing robust real-time voice pipelines using LiveKit Agents and specialized plugins for seamless user interaction.
  • Provider Benchmarking: Comparing STT and TTS providers like OpenAI, Deepgram, ElevenLabs, and Cartesia to find the best fit for specific latency or quality requirements.
  • Multilingual Voice Support: Selecting specialized engines (e.g., CLOVA or Google Cloud TTS) to optimize voice recognition and synthesis for specific languages like Korean.
  • Infrastructure Cost Optimization: Using the decision matrix to identify cost-effective providers like Deepgram Nova-3 for high-volume transcription tasks without sacrificing performance.
namevoice-realtime
descriptionReal-time voice AI knowledge including STT and TTS providers, LiveKit Agents plugins, and voice pipeline patterns. Use when working with speech-to-text, text-to-speech, voice agents, LiveKit, or any voice-related infrastructure.

Real-time Voice AI

Knowledge for building real-time voice agents with STT/TTS providers and LiveKit Agents.

Provider Selection

See stt-providers.md for speech-to-text comparison.

See tts-providers.md for text-to-speech comparison.

LiveKit Integration

See livekit-plugins.md for plugin usage and code patterns.

Quick Decision Matrix

Requirement STT TTS
Korean + Accuracy OpenAI gpt-4o-transcribe Google Cloud TTS
Korean + Low latency Deepgram Nova-3 ElevenLabs
Korean specialty CLOVA Google Cloud TTS
Cost-sensitive Deepgram Nova-3 Google Cloud TTS
Best quality OpenAI gpt-4o-transcribe ElevenLabs
Lowest latency Deepgram Nova-3 Cartesia