I just find it shocking that anyone ever used it. It’s completely useless. I ask it to do something as simple as turning on a light. Sometimes it can’t connect to the server (which is a completely stupid necessity). Sometimes it turns on the light. Sometimes it says it can’t connect to the service that turns the fucking light on. Sometimes the little lights come on to indicate it’s thinking and then just… doesn’t do anything. Sometimes it will turn on the light with just a “bing”. Sometimes it will say “okay turning on the light”.
It’s completely unpredictable at best while completing the most mundane tasks.
Sometimes it can’t connect to the server (which is a completely stupid necessity).
That’s where it does the voice processing. The only processing it does on-device is the wake word and taking commands. Actually figuring out what you mean is done in The Cloud. Doing that on-device would not only make the devices significantly more expensive, but they would also rapidly become outdated.
The rest of your complaints are valid and I’ve experienced them all myself to boot.
Doing that on-device would not only make the devices significantly more expensive, but they would also rapidly become outdated.
Uhhhh no. For one that technology already exists on home assistant and can run on a raspberry pi. For another, Android devices already do that with apps like GBoard and Google Recorder.
you’re talking about devices that are like, 5-10x the price of the nest mini.
Devices you already have and carry with you everywhere…?
I mean if it’s really about money they could at least offer a “hub” of sorts as an optional accessory that other devices can talk to for local processing for a significantly improved experience, and if people don’t want to pay for it they can just not.
I just find it shocking that anyone ever used it. It’s completely useless. I ask it to do something as simple as turning on a light. Sometimes it can’t connect to the server (which is a completely stupid necessity). Sometimes it turns on the light. Sometimes it says it can’t connect to the service that turns the fucking light on. Sometimes the little lights come on to indicate it’s thinking and then just… doesn’t do anything. Sometimes it will turn on the light with just a “bing”. Sometimes it will say “okay turning on the light”.
It’s completely unpredictable at best while completing the most mundane tasks.
That’s where it does the voice processing. The only processing it does on-device is the wake word and taking commands. Actually figuring out what you mean is done in The Cloud. Doing that on-device would not only make the devices significantly more expensive, but they would also rapidly become outdated.
The rest of your complaints are valid and I’ve experienced them all myself to boot.
Yes that’s the stupid part.
Uhhhh no. For one that technology already exists on home assistant and can run on a raspberry pi. For another, Android devices already do that with apps like GBoard and Google Recorder.
You can get a Pi that can do voice to text and NLP, maybe it can even do it with reasonable speed.
You can get a Pi that is $35.
But you won’t get a Pi that is $35 which can do voice to text and NLP with reasonable speed.
When you say “Android devices” you’re talking about devices that are like, 5-10x the price of the nest mini. Of course they’re capable.
Devices you already have and carry with you everywhere…?
I mean if it’s really about money they could at least offer a “hub” of sorts as an optional accessory that other devices can talk to for local processing for a significantly improved experience, and if people don’t want to pay for it they can just not.