Player not loading? Watch on YouTube
This tutorial continues an Echo Show conversion for Home Assistant. It assumes the device already has an open source Android installation, root access and a connection to Home Assistant. The speaker installs the View Assist Companion App APK, grants permissions and connects it to the Home Assistant integration through HACS.
The voice setup uses the Wyoming Whisper and Wyoming Piper Docker services for local speech recognition and speech synthesis. The speaker deploys Docker containers because his Home Assistant installation runs in a container; installations with add-on support can use that route instead. He connects Whisper through the Wyoming protocol on port 10300 and configures a French voice assistant. He reports about 500 MB of disk use for Whisper and says it runs well with 2 to 4 GB of RAM, based on his experience.
An optional OpenRouter connection supplies GPT-4o for general questions. Recognized home-control commands receive local handling, while the AI conversation option uses an external service. The demonstrated voice pipeline combines local speech services with hosted AI responses. Web search stays disabled, which limits the freshness of answers.
The demonstration includes a dashboard, a general-knowledge question and a heater command. The speaker stresses clear device names and reports occasional recognition failures and a pause between successive requests. He tests over a VPN and suspects it adds latency; the timings reflect that remote connection.
