Frequently Asked Questions

SH|RA is a hands-free voice assistant for Windows. You speak, your PC acts: open apps, control volume, set timers, ask AI questions, and more. Everything runs locally on your machine with no cloud dependency.

SH|RA requires Windows 10 or Windows 11. A working microphone is needed for voice input. No internet connection is required for core functionality. Only features like weather, news, Discord, and AI queries need it. Python does not need to be installed separately.

Yes. The free tier has no trial period, no credit card requirement, and no expiry. It is a permanent plan. You get the full core voice assistant with 50+ commands and offline speech recognition at no cost.

The core source files are visible in the public GitHub repository but the installer ships with PyArmor-obfuscated binaries to protect premium features. The input wrapper (pynput) is kept unobfuscated in compliance with LGPL-3.0.

Download the latest installer from the download page, run it, and follow the prompts. The installer downloads all dependencies automatically, including the voice model (~310 MB on first install). No manual setup is needed.

No. SH|RA bundles its own Python runtime. You do not need to install Python separately or manage any dependencies manually.

The first install downloads the Kokoro text-to-speech voice model (~310 MB) and installs Python package dependencies. Subsequent updates skip these downloads if the files are already present, making updates much faster.

Open the SH|RA UI, go to Settings → Utilities, and click Check for Updates. SH|RA compares your version with the latest release and prompts you to update if one is available. Your settings, memory, and macros are preserved through updates.

Use Add or Remove Programs in Windows Settings, or run uninstall.cmd from the SH|RA folder. Your config.json and memory.json files are intentionally left behind so your settings survive a reinstall.

Any working microphone works: headset, USB mic, or built-in laptop mic. A dedicated headset microphone will give you the best recognition accuracy, especially in noisy environments. SH|RA uses your Windows default microphone unless changed in Settings.

SH|RA currently supports English only. The speech recognition model (faster-whisper) supports many languages, but all commands and responses are built around English. Multi-language support may be added in a future release.

First, check that your microphone is set as the default recording device in Windows Sound Settings. Then check that SH|RA is in listening mode (status bar shows Listening). If it still isn't responding, try adjusting your microphone volume in Windows. Too low and the model won't detect speech. See the Troubleshooting guide for more steps.

Yes. SH|RA ships with 11 Kokoro voices selectable from Settings → Audio. All voices run fully offline with no internet required. The default voice is af_heart.

Yes. Select a virtual audio cable (like VB-Cable) as the Voice Output device in Settings → Audio. SH|RA's TTS will play through that device, which you can then set as your Discord microphone input. This lets your SH|RA responses play through your voice channel.

Gaming mode silences idle chatter and keeps responses focused, but still answers direct questions so you're never left hanging mid-game. If you have Steam, gaming mode activates automatically when you launch a Steam game and deactivates when you close it. You can also trigger it manually by saying "start gaming mode".

Go to Settings → Integrations and paste your API key into the API Key field. SH|RA automatically detects your provider from the key format: gsk_... for Groq, sk-ant-... for Anthropic/Claude, and sk-... for OpenAI. Groq has a free tier. Get a key at console.groq.com. See the AI Setup Guide for details.

Go to the Apps tab, enter the app name (what you'll say) and the app command (full path to the .exe or a URL), then click Add App. You can also say "add alias <shortcut> for <app name>" by voice to create a shorter name for any app that's already in the list.

Key binds work by simulating synthetic keyboard input. Most games accept this without issue. However, games protected by EasyAntiCheat or BattlEye may detect synthetic input. Behavior varies by game and anti-cheat version. Use key binds in protected games at your own risk. SH|RA and Forjem Software LLC are not responsible for any bans or flags resulting from use of this feature.

Command Macros (Premium) let you chain multiple voice commands into one phrase. For example, say "good morning" and SH|RA will open Spotify, read the weather, and list your reminders. Each step waits for the previous one to finish before starting the next. Configure them in the Integrations tab.

Paid tiers (Plus, Pro, Elite) are coming with the server infrastructure launch. We'll announce availability in the SH|RA Discord. The free tier is fully functional in the meantime.

Yes. Cancel from the user portal at any time with no penalty. You retain access to your paid tier until the end of your current billing period, then revert to the free tier automatically.

Elite tier allows you to run SH|RA on up to 3 machines under a single license. This is useful if you have a desktop and a laptop, or multiple workstations. Lower tiers are locked to one device per license.

Your config.json (settings) and memory.json (SH|RA's learned context about you) are intentionally left behind when you uninstall. Reinstalling picks them back up automatically so you don't lose your configuration.

Yes. All speech-to-text processing runs entirely on your local machine using the faster-whisper model. Your voice is never sent to any server, not Forjem's, not anyone else's. Nothing you say leaves your device unless you explicitly use a network feature like weather or AI queries.

SH|RA only makes network calls when you use features that explicitly require it: weather queries (OpenWeatherMap), news briefings, and AI queries sent directly to your chosen provider (Groq, OpenAI, or Anthropic). Discord messaging uses keyboard automation with no API calls. SH|RA also checks GitHub for updates when you click Check for Updates. No telemetry or usage data is collected.

Some anti-virus software may flag SH|RA as suspicious because it uses PyInstaller (a common packaging tool) and PyArmor (obfuscation for premium code protection). These are false positives. If your AV flags the installer, you can add an exclusion for the SH|RA install folder or check the source on the public GitHub repository.

Yes. SH|RA runs in your system tray and listens for your PTT key. It does not continuously stream audio; it only records when your push-to-talk key is held or toggled. CPU and memory usage is minimal when idle.