Silent Siren is a planned UGREEN AI NAS Skill for families of elders living alone, and for care teams in senior living and day cares. Muted motion alerts can let a 3 a.m. call for help go unheard, and many people will not send resident audio to a vendor cloud.
On request, the Skill will find audio clips, transcribe them, flag distress words such as “help me,” “I fell,” and “I can’t get up,” and note loud-sound clips. It will compare audio with camera events within two minutes. Only a finding supported by both audio and video is marked Confirmed by both. Audio without video is flagged for a person to review. Video without matching audio is listed for context and is not treated as a safety event. The caregiver will see a quote, time, snapshot, and label. Saving needs approval. The design keeps files on the NAS; testing still has to confirm no upload.
The planned workflow will use UGOS experts to locate files, transcribe audio, find keywords, and retrieve camera-event context. Reads need no approval; writes wait for approval. The Skill will use the NAS's local AI model, through Uliya. SKILL.md and three linked reference files are done; NAS testing is still needed for transcription, two-minute matching, approval waits, no-camera behavior, and confirming that the Skill never uploads.
Dignity comes first: no bathroom cameras, no diagnosis, and owner-controlled retention. It will not listen live or send alerts and is not a medical device or emergency-call substitute. Keyword and loud-sound cues may miss wordless screams or thuds; identities are never guessed. Operators own consent, signage, and recording law.
Built With
- ai
- nas
- skill.md
- ugreen
- uliya
