Offline neural TTS · Cursor & VS Code

Your editor,
out loud.

READ2ME reads the active file aloud with high-quality Piper neural voices that run 100% offline — no cloud, no API keys, nothing ever leaves your machine. It highlights every sentence and word as it speaks, so you can follow along, proofread by ear, or rest your eyes.

No internet after setup Free & open source Apple Silicon native
notes.md ● reading
1
2
3
4
5
6
7
# Release notes
 
 
 
> code block.
− 1.25x + 👤 en_US-lessac
Why READ2ME

Built for focus, not the cloud.

Everything runs on your machine. The synthesis is neural, the latency is sub-second, and your text never touches a server.

Live sentence & word highlighting

Karaoke for your code. The current sentence is anchored to its exact audio duration while the word sweep tracks along in real time — perfect for proofreading by ear or following dense docs without losing your place.

100% offline & private

No cloud, no API keys, no network needed after setup. Your words stay on your machine — always.

Status-bar controls

Play, pause, stop, and a voice picker live right in the status bar. Start from your cursor or read a selection.

Precise speed stepper

Step from 0.5× to 3× in 0.25 increments. Click the value to snap back to 1×. Your choice persists across sessions.

Markdown-aware

Syntax is stripped — no "star star" read aloud — and code blocks are skipped with a spoken marker instead of dictated.

Five neural voices, instantly switchable

US & UK English, multiple speakers. Switch voices from the picker and the read restarts from the current sentence in the new voice. Drop any Piper .onnx into the voices folder to add your own.

How it works

From file to voice in one keystroke.

01

Open a file

Click into any Markdown source in Cursor or VS Code.

02

Press the hotkey

⌃⌘R starts from your cursor. Audio begins within ~1 second.

03

Follow along

The sentence and word highlight sweep in time with the speech.

04

Stay in control

Pause, change speed, or switch voice — without leaving the editor.

Voices

Pick a voice that fits.

Five Piper neural voices ship out of the box. The picker lists any model in your voices folder, so adding more is just dropping two files in.

🇺🇸
lessacUS English · high
DEFAULT
🇺🇸
ryanUS English · high
🇺🇸
amyUS English · medium
🇬🇧
alanUK English · medium
🇬🇧
coriUK English · high

// Browse 45+ languages at rhasspy/piper-voices — drop the .onnx + .onnx.json into tts-voices/

Quick start

Up and running in three steps.

bash
# 1 · Python env
$ python3 -m venv .tts-venv
$ ./.tts-venv/bin/pip install -r tts-requirements.txt

# 2 · Download the neural voices
$ ./fetch-voices.sh

# 3 · Build + install the extension
$ cd tts-extension
$ npx --yes @vscode/vsce package --no-dependencies
$ cursor --install-extension read2me-*.vsix

Default hotkeys

Read aloud / resume⌃ ⌘ R
Pause reading⌃ ⌘ P
Stop reading⌃ ⌘ S
Change speed− / +
Select voice👤 picker

Give your docs
a voice.

Free, private, and fast. Install READ2ME and let your editor read to you.