I'm looking for an accessibility-focused voice assistant that works mostly or entirely offline and doesn't rely on a phone as its main device. Ideally, it would have a synthesized voice, programmable announcements, reminders, alarms, snooze, routines, volume controls, and a basic calendar. I'd also like to control it from a phone or computer, connect several units around the house, customize the wake word, adjust speech speed, and perhaps use a small portable version on a belt or keychain for reminders such as drinking water.
For example, I'd like to say something like, "Every Tuesday at 6 p.m., remind me to take out the trash by saying, 'The trash needs to go out tonight.'" Privacy is important, so I don't want the device constantly sending data to a company's servers. Does a system like this already exist, or could it realistically be commissioned? I'm trying to understand whether this is a practical project or an extremely expensive custom build.
3 Answers
Projects based around smart displays or “magic mirror” systems may provide some of these features, and kits could reduce the amount of custom work. However, combining offline speech recognition, natural-language commands, calendars, synchronized speakers, a custom wake word, portable hardware, and adjustable speech would still require several separate components.
A self-hosted smart-home system such as Home Assistant is probably the closest match. With local voice-processing tools, speakers, and automations, you can create reminders, routines, announcements, volume controls, and multiple-room behavior without sending everything to the cloud. The tradeoff is that setting it up would involve a fair amount of hardware selection, configuration, and troubleshooting.
The individual features are realistic, but putting them all into a polished, reliable product is more complicated and probably expensive. Phones already provide many of these functions, but they generally depend on cloud services or use the phone as the primary device, which doesn’t meet the privacy and accessibility requirements here. A custom build or accessibility-focused integrator may be more practical than commissioning an entirely new consumer product.
The main issue with using a phone assistant is precisely the cloud connection and telemetry. An offline or locally hosted setup would be much closer to what I need, even if it means accepting some limitations.
This is likely the best direction if offline operation and privacy are priorities. It can be built incrementally, starting with one speaker and a few simple reminders.