Skill UI

This advanced skill generates high-quality, natural speech from text entirely offline using the VoxCPM2 model. Supporting 30 languages, it offers three modes: standard text-to-speech, descriptive voice design, and advanced voice cloning from reference audio. All processing runs on the local device (optimized for Apple Silicon), ensuring zero API calls, zero cost, and maximum privacy. Ideal for creating professional voiceovers, podcasts, and video narration without internet dependency.

This guide provides a comprehensive project management workflow for academic submissions to the MobiSys conference. It outlines the critical path, from initial system design and on-device measurement to final paper submission and artifact preparation. The workflow helps users plan backward from the single, crucial December deadline, managing device experiment lead time and mitigating common academic failure modes.