§01 · Product

Talk. It types.
Nothing leaves your machine.

VoiceFlow is a push-to-talk dictation app for Windows. Hold a key, speak, release, and your words appear at the cursor in any app. Transcription runs entirely on your own PC, so your voice and text never touch the internet. Free to use.

§02 · Who it is for

VoiceFlow is for anyone who would rather talk than type, and especially for developers who work with an AI coding assistant all day. If you use Claude Code, Cursor, or a chat-based assistant to build software, most of your input is prose: describing a change, pasting an error, sketching a feature. Typing that is slow, and speaking it rambles. VoiceFlow closes that gap. You hold a key, say what you want, and a clean, code-aware prompt lands in the chat box, ready to send.

It also handles the everyday cases: notes, messages, email drafts, filling in a form. The same hold-to-talk flow works in any text field, in any application.

§03 · How it works

One simple loop

There is no window to switch to, no "send" button, and no account. VoiceFlow lives in the system tray and waits for the hotkey.

Two ways to trigger, set in Settings: hold (record while the key is held, transcribe on release: best for quick bursts) or toggle (tap once to start, tap again to stop: best for longer, hands-free dictation).

§04 · Output modes

Five modes, three fully offline

Every capture runs through a mode that decides how the raw transcript becomes text. Switch per capture from the tray menu, or set a default in Settings. The non-LLM modes (raw, clean, code) are fully offline and instant. The LLM modes (summary, prompt) are optional, and only run if you point VoiceFlow at a local model (such as Ollama) or a cloud endpoint you configure.

The same messy input, by mode

Both examples below are literal output from the current build.

clean  spoken:
  um so basically i need to add a
  a login function you know

output:
  So i need to add a login function.

code  spoken:
  um can you please add a dunder
  init to the fast api app

output:
  Add a __init__ to the FastAPI app.

§05 · Lead feature

The code mode and the code-aware dictionary

This is the feature built specifically for dictating to an AI coding assistant. Two things happen in code mode, both instant and on-device:

Verified examples

say: write a test for the postgres connection
get: Write a test for the PostgreSQL connection.

say: create a dot py file with a dunder init
get: Create a .py file with a __init__.

say: refactor this like the right way using type script
get: Refactor this like the right way using TypeScript.

say: add a triple equals check not equal to null
get: Add a === check != to null.

Built-in terms (v1)

You can add your own. The code dictionary merges with your personal dictionary in the config file, and your entries win. It applies in all modes, not just code, so even a clean capture gets your terms fixed. Code mode adds the politeness trimming on top. See the guide for how to add terms.

§06 · Privacy

Local-only is the default path

With on-device speech-to-text and no LLM configured, no audio and no text ever leave your machine. There is no account, no telemetry, and no upload. Speech is transcribed by a model that runs on your CPU, bundled with the app. The cleanup for raw, clean, and code modes is plain on-device text processing. The optional summary and prompt modes use an LLM only if you configure one, and you can point them at a local model to stay fully offline. Cloud speech and cloud LLM are strictly opt-in and never selected automatically.

§07 · Install

§08 · Documentation