ai-voice-production

# AI Voice Production Capability Pack > Cross-agent portable judgment for AI-assisted voice production. Covers TTS tool selection, voice cloning, long-form audiobook pipelines, narration/dubbing workflows, Apple Silicon optimization, and licensing safety. > **CONSUMES**: Text manuscripts, reference audio samples (optional), brand voice guidelines (optional). > **PRODUCES**: Production-ready audio files (WAV 48kHz preferred, 44.1kHz for ACX). Naming: `{project}/{chapter|segment}-{NNN}.wav`. > **INTERFACE**: video-creation pack defers to this pack for voice/TTS tool selection and audio quality thresholds. This pack defers to video-creation for video-specific timing (audio-to-video sync, pacing). If both packs load, this pack takes precedence for tool selection; video-creation takes precedence for visual pacing. ml-training pack provides platform selection and cost estimation for cloud GPU training. This pack defers cloud platform details to ml-training's platform-selection.md. ml-training defers voice-specific tool selection to this pack. > When both packs load for voice training: ml-training takes precedence for platform/cost decisions; ai-voice-production takes precedence for tool selection and audio quality thresholds. --- ## Step 0: Pack Prerequisites - **Python 3.10+** — most TTS tools require it - **FFmpeg** — audio post-processing and mastering - **pip or uv** — package installation in virtual environment - Memory-dependent: check `references/apple-silicon.md` for Ma...