As Snowden told us, video and audio recording capabilities of your devices are NSA spying vectors. OSS/Linux is a safeguard against such capabilities. The massive datacenter investments in US will be used to classify us all into a patriotic (for Israel)/Oligarchist social credit score, and every mega tech company can increase profits through NSA cooperation, and are legally obligated to cooperate with all government orders.
Speech to text and speech automation are useful tech, though always listening state sponsored terrorists is a non-NSA targeted path for sweeping future social credit classifications of your past life.
Some small LLMs that can be used for speech to text: https://modal.com/blog/open-source-stt
It’s not lack of software, it’s lack of hardware. Home assistant is ready as are others, but there’s no good cheap mic/speaker/esp in a box hardware
No, home assistant very much is not ready to replace an Alexa device. Home assistant mainly only does automation of smart devices, and as far as i can see from their website it does nothing else. One of the main things people use Alexa for is to play music from services like Spotify, and home assistant doesn’t appear to do that.
Home assistant has a voice assistant feature
Sorry… my experience has been trying to move my google home to something open with no cloud… it’s not been perfect for me after moving. Definitely things missing, but lots of things are better. Spotify does work with home assistant… maybe look again or send a pr
It isn’t listed anywhere on their homepage or example demos or anywhere listing its capabilities, so i did a web search to find it and I found that it sorta just kinda can do Spotify, but (1.) that isn’t listed anywhere on the home assistant abilities listing pages, which shows just how not ready for the mass market it is, and (2.) takes a ridiculous amount of very techie setup just to get it to work
https://www.home-assistant.io/integrations/spotify/
And also, out of the box can i ask it to:
-
tell me the weather?
-
set a timer?
-
set an alarm?
I don’t see anything on the website that says it can do these things. And even if it can (which doesn’t appear to be the case from their website) then the fact that the website doesn’t say it can do these things is a problem in itself that shows it isn’t ready for the mass market
Just look at the webpage for Alexa vs. Home Assistant and it’s clear that Alexa has a very wide variety of abilities and is designed to be easy to use by anyone, while the home assistant website only shows it doing smart device automation and looks like it’s not for regular folks
https://www.amazon.com/dp/B0DCCNHWV5
https://www.home-assistant.io/
I would LOVE to replace my Alexa devices with a local FOSS system, but unfortunately home assistant isn’t close to being able to do that yet
-
The HA Voice Preview is a pretty solid device, but you’re right, there isn’t really any ready made Echo/Google Home Mini replacement device - primarily because all those devices are generally sold at a loss, or at cost at best, and subsidised by your data being sold.
You won’t be able to make a Google Home Mini contender for below $50, and at that price most people will opt for the former. Good quality speakers, microphones, local processing (like the XMOS chip in the Voice Preview) all cost money, and there’s no subsidy to be made. Some older Echo devices are rootable, but the hardware tends to be somewhat exotic (meaning no open source support for specialised components), and there’s little ongoing third party support (focus has been on the display-equipped models, and to run Android on them).
All in all, “cheap” and “fully local open source voice assistant” don’t really coexist.
The issue with that is there isn’t an expensive option either. The only thing close is the home assistant voice preview and it’s still very “preview”. There’s not really any way to do it well at any price point right now.
I mean, there are many. TTS and self-hosted automation are huge in the local LLM scene.
We even have open source “omni” models now, that can ingest and output speech tokens directly (which means they get more semantic understanding from tone and such, they ‘choose’ the tone to reply with, and that it’s streamable word-by-word). They support all sorts of tool calling.
…But they aren’t easy to run. It’s still in the realm of homelabs with at least an RTX 3060 + hacky python projects.
If you’re mad, you can self-host Longcat Omni
https://huggingface.co/meituan-longcat/LongCat-Flash-Omni
And blow Alexa out of the water with a MIT-licensed model from, I kid you not, a Chinese food delivery company.
EDIT
For the curious, see:
Audio-text-to-text (and sometimes TTS): https://huggingface.co/models?pipeline_tag=audio-text-to-text&num_parameters=min%3A6B&sort=modified
TTS: https://huggingface.co/models?pipeline_tag=text-to-speech&num_parameters=min%3A6B&sort=modified
“Anything-to-anything,” generally image/video/audio/text -> text/speech: https://huggingface.co/models?pipeline_tag=any-to-any&num_parameters=min%3A6B&sort=modified
Bigger than 6B to exclude toy/test models.
I do wish there was a smaller LongCat model available. My current AI node has a hard 16GB VRAM limit (yay AMD UMA limitations), so 27B can’t really fit. An 8B dynamically loaded model would fit, and run much better.
You can do hybrid inference of Qwen 30B omni for sure. Or fully offload inference of Vibevoice Large (9B). Or really a huge array of models.
…The limiting factor is free time, TBH. Just sifting through the sea of models, seeing if they work at all, testing if quantization works and such is a huge timesink, especially if you are trying to load stuff with rocm.
And I am on ROCm - specifically on an 8945HS, which is advertised as a Ryzen AI APU yet is completely unsupported as a target with major issues around queuing and more complex models (although the new 7.0 betas have been promising but TheRock’s flip-flopping with their Docker images has been making me go crazy…).
Ah. On an 8000 APU, to be blunt, you’re likely better off with Vulkan + whatever omni models GGML supports these days. Last I checked, TG is faster and prompt processing is close to rocm.
…And yeah, that was total misadvertisement on AMD’s part. They’ve completely diluted the term kinda like TV makers did with ‘HDR’
The thing is, if AMD actually added proper support for it, given it has a somewhat powerful NPU as well… For the total TDP of the package it’s still one of the best perf per watt APU, just the damn software support isn’t there.
Feckin AMD.
The IGP is more powerful than the NPU on these things anyway. The NPU us more for ‘background’ tasks, like Teams audio processing or whatever its used for on Windows.
Yeah, in hindsight, AMD should have tasked (and still should task) a few engineers on popular projects (and pushed NPU support harder), but GGML support is good these days. It’s gonna be pretty close to RAM speed-bound for text generation.
Aye, I was actually hoping to use the NPU for TTS/STT while keeping the LLM systems GPU bound.
Home Assistant has been heavily working on that sort of functionality lately.
Home assistant continues to be fantastic, I remember it was what felt like fairly recently that all we had was OpenHAB and although it was fine, it was a bit of an uphill struggle to do anything.
There were like, about two years between OpenHAB and HA being released. Former debuted in 2011, HA saw first release in 2013.
Oh really? I could have sworn HA was a fair bit later than that
I think I used OpenHAB between about 2013 and 2018, then switched to HA around then after discovering it and reading about it for a couple of weeks.
Must have just had my head in the sand then!
