Callipso Voice Bridge

Voice-to-text for any website. Local speech recognition, no cloud. Dictate, insert, and submit in any text field.

Talk to your browser. Skip the keyboard. Callipso Voice Bridge delivers voice-to-text directly into any focused text field — email clients, messaging apps, AI chat interfaces, document editors, and thousands more. No clipboard paste, no delay. Text appears in ~15ms. Why Callipso Voice Bridge? Unlike other voice extensions that rely on cloud APIs or Chrome's built-in Web Speech API, Callipso runs local speech recognition models on your device (Apple CoreML, NVIDIA Parakeet). Your voice never leaves your machine. Key features: • Local, offline speech-to-text — no cloud, no internet required, no data sent anywhere • ~15ms text insertion via direct DOM injection (vs ~200ms clipboard paste) • Send or Submit mode — insert text only, or insert + press Enter to send messages • Works on any website with a text field — content Editable, inputs, text areas • Smart audio ducking — browser audio automatically lowers during recording • Multi-app voice routing — same voice engine routes to Chrome, terminals, and IDEs • Privacy-first — zero audio storage, zero telemetry, zero cloud dependency How it works: 1. Callipso (the desktop app) transcribes your voice locally using on-device speech recognition 2. This extension receives the text over a local WebSocket connection (localhost only) 3. Text is inserted directly into the active element — bypassing clipboard entirely 4. In Submit mode, the message is automatically sent (clicks the send button) No data leaves your machine. No analytics, no tracking, no cloud services. Permissions explained: • Host permissions: on-demand content script injection when a voice command arrives • Scripting: programmatic text insertion into the active tab • Tabs: detect the active tab for text delivery • All communication is localhost only (ws://localhost:3000) Requires Callipso desktop app (free, macOS). Download at callipso.dev