New
Point at anything on screen. Hear it.
The first thing PipeVoice does out loud. Press a hotkey, drag a box around any text — a PDF, a chat, an error dialog, a screenshot someone pasted — and it reads it to you. The text is recognised on your own PC with Windows OCR. Nothing is uploaded unless you choose a cloud voice.
A box, a window, or everything
Drag a box by default. Hold Ctrl to read the focused window, Shift for the whole screen. It works on text no app will let you select: images, locked PDFs, remote desktops.
Three voices. One is free.
Windows’ own natural voices cost nothing and never leave the machine. Prefer something warmer? Plug in your Deepgram key for Aura, or your ElevenLabs key for any voice in your library — and preview each one in Settings before you commit.
Pause, stop, start over
The pill that appears while it reads has the three buttons you actually reach for: Pause, Restart, Stop. From the keyboard, Space pauses and Esc or the hotkey stops.
Long page? Your call.
Grab a whole article and PipeVoice tells you how long the read will take, then offers a choice: read every word, or summarise it first through the polish provider you already use.
Speaks beside a screen reader
It does not fall silent when NVDA or JAWS is running. You reached for it precisely because the reader could not see that region, so it reads anyway. Quiet mode is there if you want it.
The text comes with it
What it recognised lands on your clipboard as it starts speaking, and it tells you so. Paste the paragraph into a chat, a note, or a coding agent without retyping a word from the screenshot.