What it does MMM-SUSI-AI is a MagicMirror module that adds hands-free voice interaction to a Raspberry Pi-powered smart mirror. Say “Susi,” the screen blurs, you ask your question aloud, and SUSI.AI replies via synthesized speech while the mirror returns to normal. It also supports optional face recognition for per-user authentication.

The interesting bit The project is essentially a wiring job: it stitches together Snowboy for hotword detection, Microsoft’s Bing Speech API for recognition and text-to-speech, and SUSI.AI’s own API for answers. The face recognition hook is the unusual flourish—pair it with MMM-Facial-Recognition and the mirror can identify who’s speaking, switching to that person’s SUSI account automatically.

Key highlights

• Hotword-triggered interaction with screen blur as visual feedback
• Speech pipeline uses Bing Speech API (recognition + TTS)
• Optional multi-user authentication via face recognition module
• Requires Node 6+, sox, and various native dependencies; electron-rebuild needed for Snowboy
• Explicitly alpha status; author asks for logs and screenshots with bug reports

Caveats

• Project is alpha and appears largely unmaintained; Travis CI badge points to a defunct service
• Depends on Microsoft Bing Speech API, which has been deprecated and replaced by newer Azure services
• SUSI.AI itself (accounts.susi.ai) is unclear in current status from the README alone

Verdict Worth a look if you’re already running MagicMirror and want to experiment with voice control without buying an Echo. Skip it if you need something production-stable or don’t want to debug API deprecation issues.