setup-speaklisted

Guide the user through setting up local text-to-speech for AI agents. Run this once per workstation or repository. ## Phase 1: Detect Operating System and Environment Detect the host operating system before asking any questions: - **Windows**: Use `uvx` for Pocket-TTS, Docker for Kokoro, or ONNX Runtime for Kitten TTS. - **macOS (Apple Silicon)**: Use `uvx` for Pocket-TTS, `mlx-audio` for accelerated Kokoro, or ONNX Runtime for Kitten TTS. - **macOS (Intel)**: Use `uvx` for Pocket-TTS, Docker for Kokoro, or pip for Kitten TTS. - **Linux**: Use `uvx` for Pocket-TTS, Docker for Kokoro, or pip for Kitten TTS. Confirm the detected platform to the user in one sentence and proceed to engine selection. ## Phase 2: Engine Selection Present the three local neural engines clearly. Highlight their core trade-offs: | Engine | Size | RAM Footprint | Key Strength | |---|---|---|---| | **Pocket-TTS** (Kyutai Labs) | ~100M params (150 MB) | ~300 MiB | Zero-shot voice cloning from a 5-second audio clip. Conversational cadence. MIT license. | | **Kokoro-82M** (hexgrad) | 82M params (82 MB) | ~150 MiB | 54+ studio voices across 8 languages. High prosodic fidelity. Apache 2.0 license. | | **Kitten TTS** (KittenML) | 15M to 80M (<25 MB) | ~30 MiB | Ultra-lightweight ONNX runtime. Minimal memory footprint. Apache 2.0 license. Note: Set speed to 1.25 for natural cadence. | Ask the user which engine they want to configure. ## Phase 3: Engine Installation and Verification Provide exact, cop