t2t, short for Talk to Type, is a voice-input tool whose homepage describes holding the fn key to dictate into an editable field. It records while the key is held, transcribes speech locally with Whisper and inserts the result into the field already in use. The current landing page offers a macOS download, documentation and the project's source code under the MIT license.
A separate fn-plus-control shortcut sends a transcript to a local pi command-line agent. Settings allow the user to choose a provider, model, voice output and MCP servers for that workflow. Local transcription and agent execution are different functions: a selected model provider or connected server may process information outside the device. The tool suits people who want speech input without leaving their current application, with optional access to configured agent tools rather than a standalone hosted chat interface.
