WHERE THIS TEST ENDED
Audio routing makes progress, but the dashboard appearance and spoken responses are not yet consistent.
- Echo microphone
- Fletcher on the Pi
- HDMI speech output
- Projector speaker
Move the voice to where the interface is
The preceding update ended with Fletcher still speaking through the Atom Echo. Here I return to the goal of hearing her through the projector, near the projected dashboard.
The build notes describe the split: the Echo remains the listening device, while generated speech is routed from the Pi over HDMI to the projector. A test beep checks that the output path can make sound; a real spoken reply checks that Fletcher is using that path.
Connect the voice state to the screen
The dashboard is also supposed to show the assistant when a conversation begins. I try calling Fletcher and look for her on the projected surface, but the attempts in this video do not give a clean, repeatable result.
The associated dashboard work carries voice state from the server to the interface, so the visual can distinguish listening, thinking, and speaking. Having that connection in software still needs a real conversation test at the desk.
Keep the partial result visible
The episode contains progress and failed attempts together. The projector voice path and the on-screen assistant are separate parts of the experience; success in one should not stand in for success in both.
This is a voice-and-display integration test, not proof that the everyday assistant can reliably handle every request.
About this breakdown
Matched to this original TikTok, including its spoken captions. Technical context also comes from the corresponding Fletcher build notes. The result above describes this episode; later work is identified separately.
Breakdown checked September 6, 2026.