Cursor / VS Code Extension
Text to speech for Cursor and VS Code. Turn on Agent Auto-read and finished agent replies play while you keep working. Highlight text in any file and hear it. One install, every window, zero echo. Change speed (0.75×–2×) or voice from the status bar. Search Varterm TTS in Extensions.
Long answers split into parts, so you jump forward past the preamble or back to the line that mattered instead of scrolling the chat. Markdown is stripped before speech. Same engine handles whole files, RFCs, and multi-page docs — playback starts on part one while the rest generates. Editor, selection, and clipboard work the same way. 66 free neural voices across 29 languages, or bring your own ElevenLabs key.
Cursor installs from Open VSX. VS Code can use the same search or the Marketplace. The .vsix is optional — use it only if search is empty or you want a pinned GitHub build.
Manual file is v0.1.77 from GitHub Releases. To rebuild, see the README Development section in varterm/extensions.
- Install: Extensions view → search Varterm TTS → click Install (user), not Install Workspace Extension → reload other Cursor windows.
- Optional .vsix: Command Palette →
Extensions: Install from VSIX...→ choose the package from GitHub Releases. - Run
Varterm: Connectonly if you use a custom server or token. - Click Agent Auto-read in the status bar. Finished replies from every open Cursor window join one queue and are read in the order they arrived, so parallel agents never talk over each other. Use play/pause, stop, and jump. Play reads an editor selection if you have one.
- Run
Varterm: Read Clipboard AloudorVarterm: Read Editor/Selection Aloudfor long-form docs and markdown files.