CARAPACE vs FrankenTTS

You already run coding agents on your Mac, your server, your VPS. CARAPACE puts all of them in your pocket, not as another chat app, but as a command surface: your agents stand in line, speak their results out loud, and wait for your word. TALK TO YOUR MACHINES AND LET THEM SEE WHAT YOU SEE! Vision mode intelligently uses built in Apple sensor and AI systems to feed to your agents Voice-first: speak a task, hear the reply, answer hands-free. The mic reopens when it's your turn Auto mode: every lane that finishes work gets its turn to report, like holding court Works from the lock screen and over your music. Replies duck in, music comes back Live captions of what's being spoken, right on the lock-screen card EVERY AGENT, EVERY MACHINE, ONE APP Pair unlimited machines: home Mac, office box, cloud VPS Works with the agent tools you already use: Claude Code, Codex CLI, OpenClaw, and your own custom runtimes Switch machines and lanes instantly; see who's working, who's ready, who's idle Full chat with photos and history when you'd rather type than talk YOURS, ACTUALLY CARAPACE doesn't sit between you and your AI. Your agents run on hardware you control, with your own accounts and keys. The app pairs directly with a small gateway on your machine. One line pasted into your agent sets it up, and a QR code pairs your phone. No middleman servers, no traffic through us. Setup takes about five minutes: carapace.info/setup Requires at least one machine running a supported agent CLI with your own account. CARAPACE Pro subscription: $4.99/month or $39.99/year after a 7-day free trial; auto-renews unless cancelled at least 24 hours before the period ends; manage in your App Store account settings. Terms of Use (EULA): https://www.apple.com/legal/internet-services/itunes/dev/stdeula/ Privacy Policy: https://carapace.info/privacy/

Give text a voice—entirely on your iPhone. FrankenTTS is a private, on-device text-to-speech and voice-cloning playground powered by a pure-Rust speech engine. After a one-time 1.86 GB model download, synthesis runs locally with no cloud inference. • Choose from built-in voices • Clone a voice from a short microphone recording, with a live level meter and automatic background-noise removal • Keep cloned voices in a private on-device voice library; rename, re-record, or delete them whenever you like • Share and import FrankenTTS voice cards—each carries a compact 4 KB voice fingerprint, never the original recording • Generate speech with an editable random seed • See the measured real-time factor for every run • View the waveform, play audio, and share compact M4A files • Turn generated speech into a branded MP4 video • Resume interrupted model downloads with SHA-256 verification Privacy is the default: entered text, microphone audio, voice fingerprints, imported pictures, and synthesized speech are not uploaded. Enrollment recordings are discarded after their on-device voice fingerprint is computed. Voice-card sharing and media export happen only when you explicitly choose them. Only clone or share voices you have the right and consent to use. A recent high-memory iPhone is strongly recommended.