I wonder if using the 3.5mm jack disables the internal speaker or if it could be selected via software? In theory I might want to leave it connected to an amplifier but I might only want to use the amplifier for music.
Having used Google home minis in the past I really didn't find that much benifit to voice contol aside from in a vehicle. Sure it's great when you're kneeding bread or washing dishes but I have a dish washer and don't bake my own bread so it's very rare I have an actual need for it and it false triggered all the time. Perhaps it's good for the blind or otherwise disabled who might struggle to interact with home assistant in other ways but I find for most things it's just easier to pull my phone out of my pocket. After 18 months or so the false triggering with google devices seemed to get worse and I ended up unplugging them and putting them in a draw. I've only recently pulled one of them back out but only to use as a door bell chime and for other voice announcements from homeassistant with the mic disabled to prevent any false triggering. It would be nice to do the same without a google product in the loop but I already had them sititng in a draw so it didn't make sense to try and reinvent the wheel with an ESP32 and I2S or something
The wake words that are in there are in a sense hard coded, but you have a few options. It's adding your own custom wake word that's currently not so easy and with these straight up not supported. Using Hey Jarvis seemed to work better for another RUclipsr, it's a more distinctive phrase/sound I think. But, they all work. For people who haven't set up voice pipelines and all that other stuff already there's more to do than just this basic paring, though.
Off topic, but do you still have the setup in operation where you access your weather station data via SDR radio? That was one of the first projects I followed you on, but I never got it done. I have a similar application that I'd like to try it on now. Thanks!! ~Frank
Totally different. I believe Digi did a video where he evaluated the actually hardware used in a bit more detail if i remember. Its definitely way better then seed studio but still has a laundry list of caveats. Frustrating ones of you don’t like prolonged pauses following the wake word and dislike repeating yourself or it not recognizing entities all the time.
To elaborate, if you're running home assistant on a HA yellow or green, which is nice and convenient but clearly not powerful enough to do voice well. But similar to frigate, can you do the text/speach on another machine?
@@michaeleller7514 Absolutely.. for example.. you could run ollama on a more powerful pc.. running something like llama3.2/3.3.. and use HA's built in assist pipeline to use that for both the LLM.. and also to run the whisper and piper processing.
Hi. Great videos. Two questions. 1) are the wake words customizable? 2) can you offload voice related pipelines to, say, a server? Like I’ve seen people doing with piper
“You need this!” …Hey Jarvis…..Hey Jarvis…..Hey Jarvis…. I respect your channel and the hardware may be fairly good but let’s be extremely honest assist / voice absolutely sucks. I appreciate the channels whom gave a more fair and honest evaluation while also comparing it to Alexa and google. I too want local assistants but this just isn’t there yet. We need more to go with this device, assist as well as all the local and service based LLM integrations still. We need fully customizable ways words too. Local models “trained” on or recommend as agents are all subpar too. Plus they are all fully aligned/ censored which is just lame. I’m a huge advocate for open source and local control but this thing just isn’t it until everything improves. The reviewers raving about it know their own family will get exhausted with repetitive requests etc. I’ve seen reviews where it just doesn’t compete. I may buy one anyway to support Nabu and let it sit until things improve. There’s other hardware projects that look extremely interesting. Still love the content.
I think results vary ... I've had a S3 Box for a few months and while not *perfect* has been good enough over all. More work to be done for sure but definitely far above absolutely sucking.
No. You don't NEED THIS. this is a preview version. it hardly works at the moment. this is just for the enthusiasts and the people that like to spend time with their home assistant. it is still very early. give it some time and be honest with your revies
![I.N "HALLUCINATION" | [Stray Kids : SKZ-PLAYER]](http://i.ytimg.com/vi/n5B5q1Hwt_U/mqdefault.jpg)
Do I need it? No Do I want it? Yes 😁 Think I may have to get it.Thanks for the informative video as usual.
You bet. I think it is a very nice piece of hardware when paired with HA.
I wonder if using the 3.5mm jack disables the internal speaker or if it could be selected via software? In theory I might want to leave it connected to an amplifier but I might only want to use the amplifier for music.
Just ordered mine.
Excellent!
Having used Google home minis in the past I really didn't find that much benifit to voice contol aside from in a vehicle. Sure it's great when you're kneeding bread or washing dishes but I have a dish washer and don't bake my own bread so it's very rare I have an actual need for it and it false triggered all the time. Perhaps it's good for the blind or otherwise disabled who might struggle to interact with home assistant in other ways but I find for most things it's just easier to pull my phone out of my pocket. After 18 months or so the false triggering with google devices seemed to get worse and I ended up unplugging them and putting them in a draw. I've only recently pulled one of them back out but only to use as a door bell chime and for other voice announcements from homeassistant with the mic disabled to prevent any false triggering. It would be nice to do the same without a google product in the loop but I already had them sititng in a draw so it didn't make sense to try and reinvent the wheel with an ESP32 and I2S or something
Thanks Chris!
not sure if you saw DigiBlur's WAF video on this device...... it did not go over well at all
The wake words that are in there are in a sense hard coded, but you have a few options. It's adding your own custom wake word that's currently not so easy and with these straight up not supported. Using Hey Jarvis seemed to work better for another RUclipsr, it's a more distinctive phrase/sound I think. But, they all work. For people who haven't set up voice pipelines and all that other stuff already there's more to do than just this basic paring, though.
Thanks Chris. Great video. Informative.
I was wondering does ESPHome need to be added to Home Assistant to allow this device to be added?
Can the rotary dial be used for other automations or is it locked into volume only?
Can this speaker read the answer coming from Assist?
Off topic, but do you still have the setup in operation where you access your weather station data via SDR radio? That was one of the first projects I followed you on, but I never got it done. I have a similar application that I'd like to try it on now. Thanks!! ~Frank
Awesome 😎😎
Does it work with PoE over USB-C?
Are there any differences between this and the respeaker lite from Seedstudio? They look very similar
Totally different. I believe Digi did a video where he evaluated the actually hardware used in a bit more detail if i remember. Its definitely way better then seed studio but still has a laundry list of caveats. Frustrating ones of you don’t like prolonged pauses following the wake word and dislike repeating yourself or it not recognizing entities all the time.
@@ChrisS-oo6fl i saw that video and his wife tore it apart so badly and after watching her.... I agree with her
Is there anyway to run a local LLM or similar on a more powerful local PC and have it do the voice to text etc?
To elaborate, if you're running home assistant on a HA yellow or green, which is nice and convenient but clearly not powerful enough to do voice well. But similar to frigate, can you do the text/speach on another machine?
@@michaeleller7514 Absolutely.. for example.. you could run ollama on a more powerful pc.. running something like llama3.2/3.3.. and use HA's built in assist pipeline to use that for both the LLM.. and also to run the whisper and piper processing.
Hi. Great videos. Two questions. 1) are the wake words customizable? 2) can you offload voice related pipelines to, say, a server? Like I’ve seen people doing with piper
“You need this!”
…Hey Jarvis…..Hey Jarvis…..Hey Jarvis….
I respect your channel and the hardware may be fairly good but let’s be extremely honest assist / voice absolutely sucks. I appreciate the channels whom gave a more fair and honest evaluation while also comparing it to Alexa and google. I too want local assistants but this just isn’t there yet. We need more to go with this device, assist as well as all the local and service based LLM integrations still. We need fully customizable ways words too. Local models “trained” on or recommend as agents are all subpar too. Plus they are all fully aligned/ censored which is just lame. I’m a huge advocate for open source and local control but this thing just isn’t it until everything improves. The reviewers raving about it know their own family will get exhausted with repetitive requests etc. I’ve seen reviews where it just doesn’t compete.
I may buy one anyway to support Nabu and let it sit until things improve. There’s other hardware projects that look extremely interesting. Still love the content.
I think results vary ... I've had a S3 Box for a few months and while not *perfect* has been good enough over all. More work to be done for sure but definitely far above absolutely sucking.
totally agree with you. the new box is pretty but not ready for primetime
I need this as a doorbell!
Automation can do all the things.
No. You don't NEED THIS. this is a preview version. it hardly works at the moment. this is just for the enthusiasts and the people that like to spend time with their home assistant. it is still very early. give it some time and be honest with your revies