conversational-ai-voice-engineering

Conversational voice-AI engineering team — 3 agents for production real-time voice agents: voice-ai-architect (cascade STT->LLM->TTS vs speech-to-speech, the latency budget, turn-taking / barge-in / endpointing, channel choice, build-vs-platform across Twilio / Vapi / Retell / LiveKit / Pipecat, guardrails), speech-pipeline-engineer (ASR & TTS provider choice, streaming transcription, VAD/endpointing, diarization, prosody/SSML, codecs, noise, WER), and dialog-and-integration-engineer (state, LLM orchestration and mid-call tool calling, SIP/PSTN telephony, IVR/DTMF, call routing, flow testing and eval). Ships 4 skills, a knowledge bank (4 Mermaid decision trees + a dated 2026 reference), 5 best-practices, 2 templates, 2 commands. Engineering judgment, not legal/compliance advice; the ASR/TTS/platform/telephony landscape is volatile — versions and latency numbers carry a retrieval date + [verify-at-use]; no PII, and call audio/transcripts are sensitive. Requires ravenclaude-core@>=0.7.0.