wtf is google-tts-for-nvda?

Google TTS For NVDA is an add-on for NVDA, a free screen reader used mainly by blind and low vision Windows users, that adds Google's high quality WebAssembly text to speech engine as a voice option. Instead of relying on the more robotic sounding default Windows voices, NVDA users can install this add-on and get access to Google's natural sounding voices while still running entirely offline once set up, since speech is generated locally through a background Chromium browser runtime rather than sent to a server. To use it, you need NVDA version 2024.1 or newer and one of three supported Chromium based browsers already installed on the system: Google Chrome, Microsoft Edge, or Brave. The add-on quietly runs one of these browsers in the background to render speech, Microsoft Edge additionally needs its WebView2 runtime, while Chrome and Brave do not. Because it depends on a normal interactive Windows session with that browser running, it is not meant to work in restricted environments like the Windows sign-in screen or a recovery environment. Installation happens the normal NVDA add-on way, either by opening a downloaded .nvda-addon file directly or installing it through NVDA's built in Add-on Store. After installing, you switch NVDA to use it as your synthesizer, then open the included Google TTS Voice Manager to download at least one voice language package before you can hear anything, since no voices ship installed by default. The Voice Manager lets you browse, download, and remove voice packages from an accessible dialog, and it will not let you accidentally remove the last remaining usable voice if this add-on is your active synthesizer. Beyond picking a voice, the add-on exposes normal NVDA speech settings like rate, pitch, and volume, plus its own settings panel for choosing which Chromium browser to use and configuring an optional rate boost mode that speeds up reading at the cost of extra CPU use for some voice packages. It also supports automatic per-language voice profiles, so different languages can each keep their own preferred voice and speech settings, and downloads or removals of voice packages happen in the background without blocking NVDA while you keep working. The current stable release is version 0.5.