So my girlfriend has an Alexa and really likes it but I'm not comfortable living in a house with one. How much technical knowledge would I need to make a speaker that connects to a music player app and is voice activated because that is all she uses it for.
I figure I could hook up a raspberry pi to a speaker and add a mic and voice command software that I find somewhere but I've never actually done anything like this before.
If you are not in the mood for doing much tinkering, there is a German electronics manufacturer called TechniSat and they sell a multi-function radio called DIGITRADIO 3 VOICE. It has an offline voice assistant. The voice commands can be used to pause, change audio source, increase volume, etc. The radio deck has okay-ish sound, I wouldn’t call it hi-fi but it doesn’t sound like two tin cans. The function is actually for blind people (I bought one for my grandma who is legally blind), but I guess privacy enthusiasts would find it adequate, too.
Specifically for this application, Picroft. It sounds like it might be exactly what they're looking for. Here is the why-use section. It would be interesting to hear from anybody currently using it.
Until someone with more experience finds this thread, I have seen these services recommended: Rhasspy with Home Assistant .io. Rhasspy is entirely offline, and this is Home Assistant's privacy policy.
OpenVoiceOS (OVOS) or NeonAI software (both continuations of the former Mycroft voice assistant) could be useful tools for doing a lot of the Voice Assistant tasks if you want more than just playing music. I'm not an expert on this but if you don't get another response then those are the projects I would dig into.