TTS would be a very welcome addition and I'm sure lots of others will be happy with the other languages Btw, can you satisfy my curiosity - why is the wake word Mycroft? 😅 Was that after your move there?
@@EverythingSmartHome I guess it's coming from mycroft AI (another open source assistant), Mimic can be used for Rhasspy, and it's the TTS stack in mycroft (and default wake work is... Mycroft, Sherlock's brother)
I still hope to see an ESPHome mic and speaker to send audio to a central raspberry pi such that it isn't necessary to run multiple Pi's for a multi room setup. Maybe it's already been done?
Agree!!! I bet a lot of us have ESP8266 devices scattered around the house reading temps/pressure/humidity or running lights or other things. To add a microphone and send that voice command back to Home Assistant running on the Raspberry Pi would be huge! Maybe even add a speaker at the ESP8266 or maybe more likely an ESP32 to have the Home Assistant answer back to your commands like the computer on the Enterprise in Star Trek.
@@MindForgedManacle Literally just came back to this yesterday. I'm working through what Issac Dowling has been doing here: th-cam.com/play/PL7KjlDDcKtTEYDTQ7sGCLmE0Tu7jkf6hV.html
Great Video. Please go more into detail on a)howto set this up with multiple speakers/mics in different rooms and more advances stuff that you mentioned (running services / automations)!!!1! Love your content
Thank you for this. Main reason why I gave up is the lack of “repository “ or blueprint to expand functionality without having to do it by trial and error ( that is more complicated because I should do it in Italian instead of English) I hope that the official integration will bring some love from the community
I'm glad this great project get this attention. I use it already for a few years and use it every day. Only in combination with Node-RED. I hope you also make a video with a personal wake word and how to setup multiple satellites around the house. Keep up these great videos!
@@MindForgedManacle it handles switching lights and returns me the temperature in a room for example. In Node-RED you get the spoken text as normal text which you can filter and control devices with it.
Awesome video to see pop up. As someone that highly values local control for their setup, this makes me quite excited. I haven’t had a voice assistant since I ditched Alexa when starting a fresh setup, so this gives me yet another thing to try.
If we are going to do multiple Rhaspy devices for different rooms we will want to configure them in one place all at once. They could pull a whole home configuration from a single place automatically. That configuration might also include room specific information. The server it pulls from could be aware of the location of each device and send a custom configuration for the room it is in.
This is already part of the design of Rhasspy, it is a super cool project! Very well thought out. I currently have Rhasspy running on a local server as the base station, and then have multiple raspberry pi's, one for each room, that run Rhasspy as a satellite. All the training/configuration is done on the server, and each satellite, has it's own "site id" that can be used in automations for room specific commands etc. I personally have custom scripts in Nodered that look for the "my" designator in speech(all fully customizable with keywords/slots etc in Rhasspy) to determine if commands are to be run on a room specifically, rather than a devices in another room etc.
@@BDscenes I had only briefly heard about Rhaspy "satellites" and clearly forgot about them! 🤣 This project is pretty cool! I'd love to know what you use as the main server and which PIs you use as satellites. I've been thinking it would be cool if we could create open hardware to compliment the various open assistance software projects.
@@jmr Currently I am using an old enterprise server I bought from ebay, an HP DL380 G6, running Proxmox for virtualization(The homeassistant/rhasspy base vm are running on 2 cores at 3ghz, with 8gb of ram dedicated, though the most I have seen it actually use is around 4gb), all the raspberry pis are pi4 2gb, and everything is wired, as wireless gave some performance/latency issues(may be better now, or with a different configuration)
@@BDscenes I would likely try wireless, but I too am running HA on Proxmox hosted on a NUC. I'm going to start playing around with the satellite model, but am happy to hear that someone managed to make this work. Good tip on the site id! Like you say, something to help drive my Node RED scripts.
You should add "Rhasspy" in the title, this is the "quick start guide" I had waited for a long time and I hope people would find this easily! Nice job as usual btw
Hey Lewis, Would you be able to do a follow up on making the additional speaker for another room using the Rasp Pi Zero? This is the kicker for me as I have a few smart speaker around the house. Also was wondering if a Pico W would work (probably not enough processing power or memory). Pico would be cool since I have two and can't get my hands on any more zero's.. Awesome channel!!
This is great, can't wait to give it a try. Would love to see instructions or another video for how to install on a Pi Zero. I've got a couple of those hanging around plus having voice control on every room is the goal, right?
Another Great Video - thank you!. I've just managed to setup a "satellite" in my office as my HASS is running as a virtualized server that has no Mic / Speaker. just got it to work but have had some issues on the way. 1 - PSeye works, tried with a USB Mic initially but didnt work, tried a Microsoft Web cam that was lying around (that didnt work either).. then remembered i had a PSEye in a cable box somewhere. 2 - Kaldi on the "Base" would not train and kept throwing errors , i need to re-visit it to "confirm" the words but ended up switching to "pocketspinx" and this now works. i have so far tested some lights and they work but it fails to turn my bedroom TV on even though it recognizes the words (i think this may be the way the entity is names but am still looking into that) i need to delve deeper into the "intents" but for now i'm glad it's up and running - thanks again
Hey, please expand this series about Rhasspy, i want to get rid of Alexa for a long time, but i also want in every room a esp32 as speaker/mic and just want to say ">wakewordwakeword
No way, I bought the same conference system for my Rhasspy! We're using them in our offices at work now since I was able to recommend them based on my experience with mine.
awesome! followed the steps and worked like a charm! could you provide steps to create satellite instances on pi/pi zero to have multiple mics across the house?
This is great! Do you know how this compares to other local TTS systems like Mycroft (I noticed your wakeword for Rhasspy). I'm in the middle of designing then building a house, and I want the entire house to be wired for voice/response, and I'm trying to figure out what the best options are (non-cloud only). Cheers!
Fantastic video, as usual. This is definitely going on the ever-growing to-do list. I just bought my first house so I'm finally getting to do some of the projects that I've been dreaming of for a long long time. This is definitely going to help with one of those. I mean, who doesn't want a real life Jarvis that doesn't collect your data and sell it to marketing companies?
Rhasspy has made this so much easier. Especially after I went through the trouble of generating the exact voice I wanted with machine learning. The only improvement I want on the Mic/speaker end is for the ESP Luxe to allow you to access the microphone. That and become available again 😅
I’d love to hear more about that. Does it take an actual human being speaking specific phrases or can you use recordings? It would be awesome to be able to use a really distinctive voice like Louis Armstrong or Tom Waits or even Bugs Bunny.
@@donny_bahama As long as the Mic hears my wake word spoken ("Jarvis"), it will then listen for a command phrase it recognizes and sends this to Home Assistant to process as needed. Rhasspy allows for this! Luckily I haven't been watching Marvel movies, or I assume that would trigger it too. I'll probably change it to "Hey Jarvis" if it ever becomes an issue. Really what I need is a better mic/speaker setup. An affordable Bluetooth speaker-microphone combo would allow me to place them around the house for easier command giving. Right now it's just on my computer desk.
I understand how the system works. I was referring to “generating the exact voice you wanted”. I took that to mean that you generated a custom voice for Rhasspy to use when it speaks and I was wondering how that process works.
@@donny_bahama Ah my mistake. Rhasspy is used to get the speak into text format. To get the voice I wanted, I had to train a voice model using Mozilla TTS on a bunch of clips of the voice I wanted (over 100). There's a Stack Overflow thread on the process which I followed to train the model, and having a decent GPU helps cut down the training time (though it still took quite some time). Once I had the model trained, I set up a simple program to get the responses needed to my voice commands. Since I set up all the intents in Rhasspy/Home Assistant, it was simple, if tedious.
Just wanted to say thanks for another great video. Will definitely be trying this. I love your video delivery style. Just the facts without any waffle. It makes it easy when trying to follow when setting it up. Clean, clear, concise. Thank you.
Mycroft AI or excuse me PiCroft AI as AddOn in Homeassistant... That's what I am looking for... for years actually. Thanks for charing this! You even took your time and created your own voice commands. Amazing work! I would like to see more videos about other examples for voice commands or multiple instances of PiCroft in connection with homeassistant.
I would personally make the name just the name of the room so it can be re-used for all different sensors, assuming all your sensors have a naming scheme. Ex: sensor.kitchen_temperature, sensor.living_room_humidity Otherwise (I assume) you would need a separate rooms list for each sensor.
This is a nice overview. I would love to know how you can improve the text-to-speech. A while ago Mycroft AI announced Mimic 3, which seems to be quite the gamechanger for offline voices. So it would be interesting to see if that can be integrated already.
Thanks for the hint on Mimic. It's not Windows native, and I'd rather not run a Linux VM "directly" on my windows machine, but having a 'nix box I can setup an Apache server on, send HTTP[s] requests to to make it talk sounds interesting.
Cool video didn't see this option when researching local voice assistants myself. Tried Almond (now it's called Genie) and that sucked balls. The assistant was tailored to California without the possibility to change the localization and in general, it struggled to understand what I was asking.
here's a really noob question: if I am running the speaker/mic from a different device than the Home Assistant device, do I still install both the Integration and Rhasspy itself on the HA PC? Or do I then only need the integration on the HA PC, and then Rhasspy itself on the 'other device'?
Very helpful! I can’t wait to implement this. I’d love to see how to implement multiple mics for multiple rooms. I’d also be interested in utilizing alternate TTS services for more voice options.
Answer myself (still working on the Rhaspy, so I am only assuming this works). Azure TTS has TONS of voices in a variety of languages and accents. It is very easy to implement by changing the config file.
Superb video as usual 👍 I'd defiantly be interested in more information about setting up standalone Rhasspy instances and satellites, since my Home Assistant is not in a central location.
They need to support network based speaker/mic setups. Imagine using a SIP based device that communicated back to your server (sip server) that could then take the audio stream and handle it. Short of that you need some type of usb over ethernet solution.
Thanks Lewis for making this great video 👏. Making a technical walkthrough for Rhasspy install and setup has been on my YT video TODO list for quite some time now. I think i can mark that video idea as already done now 😆.
You make such great useful tutorials... please please please make a tutorial on how to connect and use multiple Bluetooth speakers with raspberrypi.... please I have been searching how to achieve this since so long. Please consider this humble request
Many, many thanks you brought me to Rhasspy for Homeassistant! I looked long for a voice assistant that can also speak German. But it doesn't work as good as expected. The HassTurnOn, HassTurnOff and HassLightSet doesn't work with correctly set up values. Only HassGetTemperature and GetTime (and maybe others) work in German. It also look like there is not interest in updates or help anymore by the creator. Placed a Git issue and forum post 6 days ago and sent him a mail but got now response till now. That's really frustrating! It's nice that Rhasspy can tell me the temperature of Balcony and time in German but "Turned Stehlampe on/off" sounds very strange. But Rhasspy doesn't speak "turned" and "on/off" in English it speak it in German without any accent.
Awesome! Been looking for a similar solution for some time now, too. There are many options, but all have tradeoffs that are difficult to accept though.
Thank you, that is a great manual for Rhasspy! But I am still struggling following it: Since my Rhasspy profile is in german, I changed the profile file path from 'en' to 'de'. My Rhasspy is running as addon inside HassOS. Still it seems that no entities, neither lights nor switches are pulled from homeassistant. I only get error messages when I train Rhasspy. (I deleted the covers upfront, since I have none either.) Do I have to change something related to the language profile in your terminal script? Or do you have another idea?
I would like to see a new version of the video. On how to use a speakerphone in HA to create an assist sattelite after year of the voice, since lots have changed
Look forward to the video on how to add "remote" installs. My HASS is on a docker "server" under my house - so adding microphone would be useless down there lol
I understand there are some people that are concerned about sharing the "turn off the lights" command to the cloud. (Yes, I know, the mic is open to other info as well...) I don't understand why the focus in the year of the voice is not on the common voice assistants (Alexa/Google) in the devices that most tech enthusiasts own and (more or less) trust - like the Google Mini. I look forward to an integration with an UI that let's you define a voice command in HA for Google Home. Or is it hidden somewhere?
I've tried it myself last year but unfortunately the WearOs app is basically unusable and doesn't seem to be a priority at the moment, but if that ever gets supported properly I'd love to revisit. Your config is much simpler than what I had come up with too so kudos
Thanks for your video, it is very ussefull, I have a question is there any chance to use the smart phone to send a voice comand, instead of getting a specific hard with mic and speakers , if so , can you give some ideas on how to implement it. if possible everyone in my family ca acces to home assistant with phone and send a voice commands. Thanks in advance.
Hi Lewis. Thanks for all your videos, which are my book of receipes. I carefully followed your instructions and installed the RHASSPY full version (up-to-date HA OS on Rpi4). Voice recognition works fine and speach to text, also. As you mentioned, I got the "cover" error at the end, and I deleted the lines related to covers. But once I have done that, then the same error occurred for "lights" (AssertionError: No output from ['/share/rhasspy/profiles/en/slot_programs/hass/entities', 'light']), whereas I do have lights in my HA installation ("light.techo_comedor" for example). Would you have any idea of what I have done wrong? Thanks again!
Perhaps a compact Rhasspy satellite will be the next hardware project for Lewis and his Dad. Would love to have something small with an integrated mircrophone.
Great video! Thanks! But can't get the slot_program for the downloading off entities to work, just getting a failure off missing slots. What am I missing or doing wrong?
I followed the directions to the letter, but when i try to train Rhasspy i get "AssertionError: Missing slot $hass/entities,light", i am not sure what i am doing wrong. I think the problem is because i am running the HAOS version of HA and probably broke something, i put api: in the config file + intent: but no luck. What i did do is dig a bit deeper into the Rhasspy documentation and put the script into Rhasspy itself.. under your profile you make a slot_programs/hass directory and put the script there.. then everything works like a charm. One other thing i had to do was change the script to have an absolute path to my profile.json vs using the environment variable (which wasn't set). Also, and this is very important, make sure all your HA devices have nice friendly names, otherwise Rhasspy won't train properly. Maybe i will someday redo my overall setup and start fresh, maybe.
Great video, thanks for the detailed explanation. I'm curious if Rhasspy can "follow a conversation". Meaning for example that you ask about the temperature in the kitchen and after it responds one follows up with "and in the bedroom?".
Can you make Amazon echo speakers and mics work with this? Also can this be used to make announcements over the speakers, for example. When the front door contact is separated the speakers say “Front door open” on all the speakers.
Ok Tech Community, I need some help! I'm not really tech savvy and I'm trying to find the safest cheapest alternative to replace somebody's Echo device because of privacy issues with them. (said person refuses to do without a smart speaker...) Is it easy to do for a beginner? What should I know before buying a raspberry pi?(I just want the cheapest option that will get the job done) How much money will every thing cost just to make it work for mostly simple tasks like playing rain noise for sleep or telling the time or weather? (I know said persons smart phone can do all that but I'm dealing with a stubborn one... so this is where I'm at unfortunately...) please any help is appreciated.
The main way to do this in multiple rooms is to have several different Rhasspy installations on raspberry pi computers (satellites) in different rooms and configure them to use the same master (usually installed on the same server hardware as Home Assistant). The satellites will have different names and you can use those names to send the TTS information to that room.
This is a great guide, really makes me feel like I could do this myself! One thing I'd like a little more information about would be good hardware to use with this. I understand I can run this on probably anything that runs linux, but I have no idea what kind of speaker/microphone I should be looking for to get a good result. If I can't get a response similar to how I talk to my Google speaker facing a random direction, then I'd probably get frustrated with it.
First of all, very nice video. The question I have is how functional is this mode of voice commands since they depend on the position of a microphone. As you understand, I mean that if I place the microphone in the living room and I'm in the kitchen wanting to give a voice command, it's impossible. On the other hand, it is not so easy to fill the house with microphones. My thought is if I can use my cell phone as the voice gate for the home assistant voice commands. Thank you very much for your time.
THX for the Video. I have now installed my rhasspy on a pi zero and need to know how to connect external rhasspy items to ha. Maybe you can create a video therefore. many greetings, gerald
Hi there, can you make a video on how to setup Seven Segments OCR with Home assistant? There is very little guide on this and absolutely no videos till now.
@@EverythingSmartHome I guess it's looking good nowadays :)
2 ปีที่แล้ว +1
In hopes I can get it working fast I bought the same Anker, but sadly, the mic is not available in HA, I tried tons of various things but cant get it working :(
Thanks for this video. I have to questions: 1. would it be possible to connect the speaker/ microfon combination via BT to the RASPI with HA? Would help a lot to place the mic in a better position. 2. Can you make a video to have a second Rahsspy installed on a Pi Zero connected via W-Lan?
A Bluetooth speaker/mix should work fine, as long as the distance/number of walls to the speaker isn't too great. ...and if you find a good Bluetooth speaker-microphone combo, let us know!
@@MindForgedManacle I tried to find any information how to pair, but i find nothing. I have installed BT in Homeassistant and the one device (the BT device of the Raspi) is found. The next step is missed by me. How can i achieve it that the Speaker combo is paired so i can use it
Thanks for this video. I've been using Alexa for the last 5 years now in combination with 1Home and Loxone. It seems 1Home is slowly loosing the competitive battle with their solution for now resulting in my setup no longer working. I have a Sonos speaker in every room setup with Alexa and until a few weeks ago that worked fine but it's not offline at all. I'm looking for an offline voice assistant too. The Elephant in the room here is what you mentioned early in the video. Do I need more than 1 HA which I think would be very inconvenient. Can you elaborate on that?
Great explanation! Thanks. Do you have also a video how to setup another rhasspy device? For example I use that one with the device, where Home Assistant runs on, but now I want in another room voice control too. How do I setup and how is it than connected with Home Assistant? Is it possible to play mp3 files / playlists over them?
My Rhasspy junior configuration only shows the 'Default input'. I have tried many microphones in the Raspberry Pi's USB ports but none will show up in the Config options pull down menu.
Great Video, you've made some great instructional videos! This is like its 2001 all over again. Where does Rhasspy get the Brewer's scores? How does he know the weather in Ketchican AK? I don't want a deaf butler answering the door.
Great Video! I’ve bought the same Anker device! I’m running home assistant at Raspberry pi, but i have problems with the microphones of the Anker device. They are not running... I choose the Anker device in audio configuration and i can listen sounds but, micros aren’t listen anything. Had you got the same issues? Any idea hoy to make them work? Thanks!
Very interesting concept. Is there a way for it to execute an action in the room you are currently located, in case you don't mention a room in your command, unless you specify which room you want the command to be executed? That would be very nice!
Great video mate am currently looking the same thing but addding a mini mic and speak to every room in the house , my current project is trying to figure out some type of door switch that is outdoor rated to alert me to gates been open and closed any suggestions?
Why did you not show you using Rhasspy at the start of the video? How do I know I want to install this voice assistant if I don't even know if I want it.
So then it's not really "local" voice control... If I have to bring a RPi into each room to make it work. Unless of course i live (and work) inside the server room... Or bring the server room into my workspace. Would be nice to be able to connect the speaker/mic remotely to home assistant on the RPi.
Great video, thanks for giving Rhasspy a try! It definitely does need refinement, and some updates. Glad to see Junior is working well at least!
Thanks so much for making it, its really great and much needed! Junior is so incredibly simple to setup, very impressive!
@@EverythingSmartHome You're welcome! My next steps for Junior will be to add TTS and some non-English models.
TTS would be a very welcome addition and I'm sure lots of others will be happy with the other languages
Btw, can you satisfy my curiosity - why is the wake word Mycroft? 😅 Was that after your move there?
@@EverythingSmartHome I suspect it's for the computer in The Moon is a Harsh Mistress by Heinlein.. If it is, kudos to @synesthesiam for that..
@@EverythingSmartHome I guess it's coming from mycroft AI (another open source assistant), Mimic can be used for Rhasspy, and it's the TTS stack in mycroft (and default wake work is... Mycroft, Sherlock's brother)
Cloud-based voice control can go IN THE BIN!
I agree!!
I hear you! Just saw the Cambridge Analytics on Netflix. I'm getting paranoid.
Absolutely chopped
@@rudiklein what show did you watch on Netflix?
@@ntatek The great hack. I now sleep with my eyes wide open 😳. I thought the internet was always truthful. 🤣
I still hope to see an ESPHome mic and speaker to send audio to a central raspberry pi such that it isn't necessary to run multiple Pi's for a multi room setup.
Maybe it's already been done?
That would be a game changer for sure.
Agree!!! I bet a lot of us have ESP8266 devices scattered around the house reading temps/pressure/humidity or running lights or other things. To add a microphone and send that voice command back to Home Assistant running on the Raspberry Pi would be huge! Maybe even add a speaker at the ESP8266 or maybe more likely an ESP32 to have the Home Assistant answer back to your commands like the computer on the Enterprise in Star Trek.
I’m working on this exact thing. Looking for collaborators.
@@devinhedge Any update on that, or a place to follow your work?
@@MindForgedManacle Literally just came back to this yesterday. I'm working through what Issac Dowling has been doing here: th-cam.com/play/PL7KjlDDcKtTEYDTQ7sGCLmE0Tu7jkf6hV.html
Great Video. Please go more into detail on a)howto set this up with multiple speakers/mics in different rooms and more advances stuff that you mentioned (running services / automations)!!!1! Love your content
4:34 was the best moment of the video! Great overview of Rhasspy!
😂 so happy with myself!
Brilliant Lewis. You got my attention at the point of telling that you can run scripts and automations using this. Now I need this in my life...
Nice, look forward to seeing how it works for you 😅
Thank you for this. Main reason why I gave up is the lack of “repository “ or blueprint to expand functionality without having to do it by trial and error ( that is more complicated because I should do it in Italian instead of English)
I hope that the official integration will bring some love from the community
I'm glad this great project get this attention. I use it already for a few years and use it every day. Only in combination with Node-RED.
I hope you also make a video with a personal wake word and how to setup multiple satellites around the house. Keep up these great videos!
Thanks, appreciate that!
What does Node RED handle in your setup with respect to voice commands?
@@MindForgedManacle it handles switching lights and returns me the temperature in a room for example.
In Node-RED you get the spoken text as normal text which you can filter and control devices with it.
Awesome video to see pop up. As someone that highly values local control for their setup, this makes me quite excited. I haven’t had a voice assistant since I ditched Alexa when starting a fresh setup, so this gives me yet another thing to try.
Nice, let me know what you think!
If we are going to do multiple Rhaspy devices for different rooms we will want to configure them in one place all at once. They could pull a whole home configuration from a single place automatically. That configuration might also include room specific information. The server it pulls from could be aware of the location of each device and send a custom configuration for the room it is in.
This is already part of the design of Rhasspy, it is a super cool project! Very well thought out. I currently have Rhasspy running on a local server as the base station, and then have multiple raspberry pi's, one for each room, that run Rhasspy as a satellite. All the training/configuration is done on the server, and each satellite, has it's own "site id" that can be used in automations for room specific commands etc. I personally have custom scripts in Nodered that look for the "my" designator in speech(all fully customizable with keywords/slots etc in Rhasspy) to determine if commands are to be run on a room specifically, rather than a devices in another room etc.
@@BDscenes I had only briefly heard about Rhaspy "satellites" and clearly forgot about them! 🤣 This project is pretty cool! I'd love to know what you use as the main server and which PIs you use as satellites. I've been thinking it would be cool if we could create open hardware to compliment the various open assistance software projects.
@@jmr Currently I am using an old enterprise server I bought from ebay, an HP DL380 G6, running Proxmox for virtualization(The homeassistant/rhasspy base vm are running on 2 cores at 3ghz, with 8gb of ram dedicated, though the most I have seen it actually use is around 4gb), all the raspberry pis are pi4 2gb, and everything is wired, as wireless gave some performance/latency issues(may be better now, or with a different configuration)
@@BDscenes Thanks! I much prefer wired myself whenever possible.
@@BDscenes I would likely try wireless, but I too am running HA on Proxmox hosted on a NUC. I'm going to start playing around with the satellite model, but am happy to hear that someone managed to make this work. Good tip on the site id! Like you say, something to help drive my Node RED scripts.
You should add "Rhasspy" in the title, this is the "quick start guide" I had waited for a long time and I hope people would find this easily!
Nice job as usual btw
Thanks!
Hey Lewis, Would you be able to do a follow up on making the additional speaker for another room using the Rasp Pi Zero? This is the kicker for me as I have a few smart speaker around the house. Also was wondering if a Pico W would work (probably not enough processing power or memory). Pico would be cool since I have two and can't get my hands on any more zero's.. Awesome channel!!
This is great, can't wait to give it a try. Would love to see instructions or another video for how to install on a Pi Zero. I've got a couple of those hanging around plus having voice control on every room is the goal, right?
I was just able to get it work on a pi Zero 2 w with DietPi, but Its a bit slow in docker, I am wondering if dietpi is the best baseos for this
Another Great Video - thank you!. I've just managed to setup a "satellite" in my office as my HASS is running as a virtualized server that has no Mic / Speaker. just got it to work but have had some issues on the way.
1 - PSeye works, tried with a USB Mic initially but didnt work, tried a Microsoft Web cam that was lying around (that didnt work either).. then remembered i had a PSEye in a cable box somewhere.
2 - Kaldi on the "Base" would not train and kept throwing errors , i need to re-visit it to "confirm" the words but ended up switching to "pocketspinx" and this now works.
i have so far tested some lights and they work but it fails to turn my bedroom TV on even though it recognizes the words (i think this may be the way the entity is names but am still looking into that)
i need to delve deeper into the "intents" but for now i'm glad it's up and running - thanks again
Hey, please expand this series about Rhasspy, i want to get rid of Alexa for a long time, but i also want in every room a esp32 as speaker/mic and just want to say ">wakewordwakeword
No way, I bought the same conference system for my Rhasspy! We're using them in our offices at work now since I was able to recommend them based on my experience with mine.
awesome! followed the steps and worked like a charm! could you provide steps to create satellite instances on pi/pi zero to have multiple mics across the house?
Reading through the doc, deployment using remote satellites looks very intriguing.
As always, thanks for the great video Lewis!
Would love to see a tut on satellites!
This is great! Do you know how this compares to other local TTS systems like Mycroft (I noticed your wakeword for Rhasspy). I'm in the middle of designing then building a house, and I want the entire house to be wired for voice/response, and I'm trying to figure out what the best options are (non-cloud only). Cheers!
Fantastic video, as usual. This is definitely going on the ever-growing to-do list. I just bought my first house so I'm finally getting to do some of the projects that I've been dreaming of for a long long time. This is definitely going to help with one of those. I mean, who doesn't want a real life Jarvis that doesn't collect your data and sell it to marketing companies?
Haha sorry for adding to the list! 😅
YEEEEESSS! I was fiddling around with Rhasspy but never got it to run. Thanks for the great video mate!!!
Extremely cool! And I don't think it's too technical, thanks for going in depth!
Rhasspy has made this so much easier. Especially after I went through the trouble of generating the exact voice I wanted with machine learning. The only improvement I want on the Mic/speaker end is for the ESP Luxe to allow you to access the microphone. That and become available again 😅
I was gonna ask if the raspiaudio devices he covered recently could be used for rasspy, too.
I’d love to hear more about that. Does it take an actual human being speaking specific phrases or can you use recordings? It would be awesome to be able to use a really distinctive voice like Louis Armstrong or Tom Waits or even Bugs Bunny.
@@donny_bahama As long as the Mic hears my wake word spoken ("Jarvis"), it will then listen for a command phrase it recognizes and sends this to Home Assistant to process as needed. Rhasspy allows for this! Luckily I haven't been watching Marvel movies, or I assume that would trigger it too. I'll probably change it to "Hey Jarvis" if it ever becomes an issue.
Really what I need is a better mic/speaker setup. An affordable Bluetooth speaker-microphone combo would allow me to place them around the house for easier command giving. Right now it's just on my computer desk.
I understand how the system works. I was referring to “generating the exact voice you wanted”. I took that to mean that you generated a custom voice for Rhasspy to use when it speaks and I was wondering how that process works.
@@donny_bahama Ah my mistake. Rhasspy is used to get the speak into text format. To get the voice I wanted, I had to train a voice model using Mozilla TTS on a bunch of clips of the voice I wanted (over 100). There's a Stack Overflow thread on the process which I followed to train the model, and having a decent GPU helps cut down the training time (though it still took quite some time).
Once I had the model trained, I set up a simple program to get the responses needed to my voice commands. Since I set up all the intents in Rhasspy/Home Assistant, it was simple, if tedious.
Just wanted to say thanks for another great video. Will definitely be trying this. I love your video delivery style. Just the facts without any waffle. It makes it easy when trying to follow when setting it up. Clean, clear, concise. Thank you.
Mycroft AI or excuse me PiCroft AI as AddOn in Homeassistant... That's what I am looking for... for years actually.
Thanks for charing this! You even took your time and created your own voice commands. Amazing work!
I would like to see more videos about other examples for voice commands or multiple instances of PiCroft in connection with homeassistant.
I would personally make the name just the name of the room so it can be re-used for all different sensors, assuming all your sensors have a naming scheme. Ex: sensor.kitchen_temperature, sensor.living_room_humidity
Otherwise (I assume) you would need a separate rooms list for each sensor.
This is a nice overview. I would love to know how you can improve the text-to-speech.
A while ago Mycroft AI announced Mimic 3, which seems to be quite the gamechanger for offline voices.
So it would be interesting to see if that can be integrated already.
Agree, Mimic 3 looks amazing
@@EverythingSmartHome Had to look myself and it seems it actually already is inside. So should not be hard to use that.
I do plan to get Mimic 3 integrated into Rhasspy soon. I wrote both, so it should be easy :)
Thanks for the hint on Mimic. It's not Windows native, and I'd rather not run a Linux VM "directly" on my windows machine, but having a 'nix box I can setup an Apache server on, send HTTP[s] requests to to make it talk sounds interesting.
@@synesthesiam Mike, you are by far, one of the most enthusiast open voice people i know - for sure.
Cool video didn't see this option when researching local voice assistants myself. Tried Almond (now it's called Genie) and that sucked balls. The assistant was tailored to California without the possibility to change the localization and in general, it struggled to understand what I was asking.
here's a really noob question:
if I am running the speaker/mic from a different device than the Home Assistant device, do I still install both the Integration and Rhasspy itself on the HA PC? Or do I then only need the integration on the HA PC, and then Rhasspy itself on the 'other device'?
Very helpful! I can’t wait to implement this. I’d love to see how to implement multiple mics for multiple rooms. I’d also be interested in utilizing alternate TTS services for more voice options.
Answer myself (still working on the Rhaspy, so I am only assuming this works). Azure TTS has TONS of voices in a variety of languages and accents. It is very easy to implement by changing the config file.
Superb video as usual 👍
I'd defiantly be interested in more information about setting up standalone Rhasspy instances and satellites, since my Home Assistant is not in a central location.
They need to support network based speaker/mic setups. Imagine using a SIP based device that communicated back to your server (sip server) that could then take the audio stream and handle it. Short of that you need some type of usb over ethernet solution.
Thanks Lewis for making this great video 👏. Making a technical walkthrough for Rhasspy install and setup has been on my YT video TODO list for quite some time now. I think i can mark that video idea as already done now 😆.
Haha thanks, appreciate it! Keep doing the good work your doing!
Great, detailed and clear tutorial 👌🏼 thank you 👏🏼 do you know if it is possible to use a Bluetooth connected speaker/mike ?
Excellent! It's much better than just a year ago. Still fairly convoluted to set up, but getting there.
Thanks! Agree it's really good but could be a little easier for newbies!
You make such great useful tutorials... please please please make a tutorial on how to connect and use multiple Bluetooth speakers with raspberrypi.... please I have been searching how to achieve this since so long. Please consider this humble request
Many, many thanks you brought me to Rhasspy for Homeassistant! I looked long for a voice assistant that can also speak German. But it doesn't work as good as expected. The HassTurnOn, HassTurnOff and HassLightSet doesn't work with correctly set up values. Only HassGetTemperature and GetTime (and maybe others) work in German. It also look like there is not interest in updates or help anymore by the creator. Placed a Git issue and forum post 6 days ago and sent him a mail but got now response till now. That's really frustrating! It's nice that Rhasspy can tell me the temperature of Balcony and time in German but "Turned Stehlampe on/off" sounds very strange. But Rhasspy doesn't speak "turned" and "on/off" in English it speak it in German without any accent.
thanks I tried Rhasspy about a year ago without the integration and it wasn't quite there, going to give this another go next week
Great video, would really love to see more videos on this to see what could be done with it.
Awesome! Been looking for a similar solution for some time now, too.
There are many options, but all have tradeoffs that are difficult to accept though.
Thanks!
Indeed, I think we will see some cool stuff over the coming year...I hope anyways!
Thank you, that is a great manual for Rhasspy!
But I am still struggling following it: Since my Rhasspy profile is in german, I changed the profile file path from 'en' to 'de'.
My Rhasspy is running as addon inside HassOS.
Still it seems that no entities, neither lights nor switches are pulled from homeassistant. I only get error messages when I train Rhasspy. (I deleted the covers upfront, since I have none either.)
Do I have to change something related to the language profile in your terminal script? Or do you have another idea?
I would like to see a new version of the video. On how to use a speakerphone in HA to create an assist sattelite after year of the voice, since lots have changed
Look forward to the video on how to add "remote" installs. My HASS is on a docker "server" under my house - so adding microphone would be useless down there lol
Rhasspy has also satellites. They are dumb and just forward the sound(mic and speaker) thru/from the Rhasspy server
Cool! I'm going to give this a try. I'd love to see another video when you've learned/discovered more features :)
I understand there are some people that are concerned about sharing the "turn off the lights" command to the cloud. (Yes, I know, the mic is open to other info as well...)
I don't understand why the focus in the year of the voice is not on the common voice assistants (Alexa/Google) in the devices that most tech enthusiasts own and (more or less) trust - like the Google Mini. I look forward to an integration with an UI that let's you define a voice command in HA for Google Home. Or is it hidden somewhere?
I'm not sure what the quality is like on them, but there's other usb conference speaker/mic combos for
I've tried it myself last year but unfortunately the WearOs app is basically unusable and doesn't seem to be a priority at the moment, but if that ever gets supported properly I'd love to revisit. Your config is much simpler than what I had come up with too so kudos
Thanks for your video, it is very ussefull, I have a question is there any chance to use the smart phone to send a voice comand, instead of getting a specific hard with mic and speakers , if so , can you give some ideas on how to implement it. if possible everyone in my family ca acces to home assistant with phone and send a voice commands. Thanks in advance.
Cool! I wonder how the custom commands works with another languages... I might need to test it some day! 😀
Hi Lewis. Thanks for all your videos, which are my book of receipes. I carefully followed your instructions and installed the RHASSPY full version (up-to-date HA OS on Rpi4). Voice recognition works fine and speach to text, also. As you mentioned, I got the "cover" error at the end, and I deleted the lines related to covers. But once I have done that, then the same error occurred for "lights" (AssertionError: No output from ['/share/rhasspy/profiles/en/slot_programs/hass/entities', 'light']), whereas I do have lights in my HA installation ("light.techo_comedor" for example). Would you have any idea of what I have done wrong? Thanks again!
Perhaps a compact Rhasspy satellite will be the next hardware project for Lewis and his Dad.
Would love to have something small with an integrated mircrophone.
Great video! Thanks!
But can't get the slot_program for the downloading off entities to work, just getting a failure off missing slots. What am I missing or doing wrong?
I'd really like a tutorial on how to make an esp32 satellite :D
I followed the directions to the letter, but when i try to train Rhasspy i get "AssertionError: Missing slot $hass/entities,light", i am not sure what i am doing wrong. I think the problem is because i am running the HAOS version of HA and probably broke something, i put api: in the config file + intent: but no luck. What i did do is dig a bit deeper into the Rhasspy documentation and put the script into Rhasspy itself.. under your profile you make a slot_programs/hass directory and put the script there.. then everything works like a charm. One other thing i had to do was change the script to have an absolute path to my profile.json vs using the environment variable (which wasn't set). Also, and this is very important, make sure all your HA devices have nice friendly names, otherwise Rhasspy won't train properly. Maybe i will someday redo my overall setup and start fresh, maybe.
I hope to see more languages in the future.
Could this be combined with the ESPMuse you showcased some time ago? Or their HAT that includes mic+speaker?
Amazing. How can we enable a custom wake word??
That is awesome! Now if we could just use that speaker as a distributed home audio setup for music 🤔
Great video, thanks for the detailed explanation.
I'm curious if Rhasspy can "follow a conversation". Meaning for example that you ask about the temperature in the kitchen and after it responds one follows up with "and in the bedroom?".
Can you make Amazon echo speakers and mics work with this? Also can this be used to make announcements over the speakers, for example. When the front door contact is separated the speakers say “Front door open” on all the speakers.
I have long wondered whether it’s possible to hack and reprogram an old Echo.
Ok Tech Community, I need some help! I'm not really tech savvy and I'm trying to find the safest cheapest alternative to replace somebody's Echo device because of privacy issues with them. (said person refuses to do without a smart speaker...) Is it easy to do for a beginner? What should I know before buying a raspberry pi?(I just want the cheapest option that will get the job done) How much money will every thing cost just to make it work for mostly simple tasks like playing rain noise for sleep or telling the time or weather? (I know said persons smart phone can do all that but I'm dealing with a stubborn one... so this is where I'm at unfortunately...) please any help is appreciated.
Did I miss the spot where you explained how to add/select a microphone/speaker?
I do miss some features here, does the speaker know in which room he is? What happens with multiple speakers?
The main way to do this in multiple rooms is to have several different Rhasspy installations on raspberry pi computers (satellites) in different rooms and configure them to use the same master (usually installed on the same server hardware as Home Assistant). The satellites will have different names and you can use those names to send the TTS information to that room.
This is a great guide, really makes me feel like I could do this myself!
One thing I'd like a little more information about would be good hardware to use with this. I understand I can run this on probably anything that runs linux, but I have no idea what kind of speaker/microphone I should be looking for to get a good result. If I can't get a response similar to how I talk to my Google speaker facing a random direction, then I'd probably get frustrated with it.
I bought a second hand USB conference Jabra speaker. I have had it for about 3 weeks and it has been fine.
can i use echo dot as a speaker and mic?
First of all, very nice video. The question I have is how functional is this mode of voice commands since they depend on the position of a microphone. As you understand, I mean that if I place the microphone in the living room and I'm in the kitchen wanting to give a voice command, it's impossible. On the other hand, it is not so easy to fill the house with microphones. My thought is if I can use my cell phone as the voice gate for the home assistant voice commands. Thank you very much for your time.
Excellent video!
Great work man. 👏
THX for the Video. I have now installed my rhasspy on a pi zero and need to know how to connect external rhasspy items to ha. Maybe you can create a video therefore. many greetings, gerald
Great video, as always Lewis. I can't wait to give this a go
Thank you, enjoy!
Very cool, thanks for the quick guide 👍
i'd love to see a video about integrating esp muse with rhasspy!!
Hi there, can you make a video on how to setup Seven Segments OCR with Home assistant? There is very little guide on this and absolutely no videos till now.
What is the repository link? The github link above seems not to work for me.
That is amazing, I really want it for myself now!
Btw, what shell is that on powershell? Can you share your configuration?
Just the regular windows terminal in Windows 11
@@EverythingSmartHome I guess it's looking good nowadays :)
In hopes I can get it working fast I bought the same Anker, but sadly, the mic is not available in HA, I tried tons of various things but cant get it working :(
I had the same idea, but sadly, it seems that Ankers firmware does not like Linux :/
Thanks for this video. I have to questions:
1. would it be possible to connect the speaker/ microfon combination via BT to the RASPI with HA? Would help a lot to place the mic in a better position.
2. Can you make a video to have a second Rahsspy installed on a Pi Zero connected via W-Lan?
A Bluetooth speaker/mix should work fine, as long as the distance/number of walls to the speaker isn't too great.
...and if you find a good Bluetooth speaker-microphone combo, let us know!
@@nitramremoeb4674 It wasn't able to pair like a normal Bluetooth device?
@@MindForgedManacle I tried to find any information how to pair, but i find nothing. I have installed BT in Homeassistant and the one device (the BT device of the Raspi) is found. The next step is missed by me. How can i achieve it that the Speaker combo is paired so i can use it
Now if I can have firmware install this on my Google homes it would be awesome
Thanks for this video.
I've been using Alexa for the last 5 years now in combination with 1Home and Loxone. It seems 1Home is slowly loosing the competitive battle with their solution for now resulting in my setup no longer working. I have a Sonos speaker in every room setup with Alexa and until a few weeks ago that worked fine but it's not offline at all.
I'm looking for an offline voice assistant too. The Elephant in the room here is what you mentioned early in the video. Do I need more than 1 HA which I think would be very inconvenient. Can you elaborate on that?
Nice. Shouldnt it be possible to use the TTS engine provided by HACloud/NabuCasa instead? because it sounds very natural and pleasant
Maybe! Not something I've looked into yet!
You could try setting up your speaker to be detected as a media player on your network, then you can play TTS audio to it from home assistant.
Great video, very informative. Going to build this right now.
Brilliant! You just earned a like and subscribe, this is precisely what I was looking for.
Thanks! 🙏🏻
Great explanation! Thanks. Do you have also a video how to setup another rhasspy device? For example I use that one with the device, where Home Assistant runs on, but now I want in another room voice control too. How do I setup and how is it than connected with Home Assistant? Is it possible to play mp3 files / playlists over them?
I love the idea of these but I still can't get past how robotic the sound. Are there any options for a more natural sounding voice model?
This was amazing! Finally reason to install Home Assistant :)
There is so many good reason though 👀😅
My Rhasspy junior configuration only shows the 'Default input'. I have tried many microphones in the Raspberry Pi's USB ports but none will show up in the Config options pull down menu.
Great Video, you've made some great instructional videos! This is like its 2001 all over again. Where does Rhasspy get the Brewer's scores? How does he know the weather in Ketchican AK? I don't want a deaf butler answering the door.
Thanks! Haha the voice certainly could do with some improvement!
Great Video! I’ve bought the same Anker device! I’m running home assistant at Raspberry pi, but i have problems with the microphones of the Anker device. They are not running... I choose the Anker device in audio configuration and i can listen sounds but, micros aren’t listen anything. Had you got the same issues? Any idea hoy to make them work? Thanks!
just what i’ve been looking for! can you post the code for your v2 presence sensor? Mine is working but having issues with random flapping
Join the discord I'm sure we can get you sorted out
great clarification! can we use rhasspy offline without a laptop?
Great video !
Do you know of any way to use your phone's mic and speaker instead of a plugged in mic/speaker combo ?
Very interesting concept.
Is there a way for it to execute an action in the room you are currently located, in case you don't mention a room in your command, unless you specify which room you want the command to be executed?
That would be very nice!
You could do that with a custom sentence as shown in the video, since you have to setup sentences on each one - hope that makes sense
Great video mate am currently looking the same thing but addding a mini mic and speak to every room in the house , my current project is trying to figure out some type of door switch that is outdoor rated to alert me to gates been open and closed any suggestions?
Perhaps there's an install method for those running a NUC with HA operating system? Because even after restart it doesn't see the microphone..
Why did you not show you using Rhasspy at the start of the video? How do I know I want to install this voice assistant if I don't even know if I want it.
Could you make your own voice for this? I really like the idea of it being David Attenborough for example
Is it possible to make this one work with ChatGPT? I have it working with an ESP S3 but the volume is very low.
it's useful if you split the video description with the chapters
So then it's not really "local" voice control... If I have to bring a RPi into each room to make it work. Unless of course i live (and work) inside the server room... Or bring the server room into my workspace.
Would be nice to be able to connect the speaker/mic remotely to home assistant on the RPi.
I purchased the very same conference speaker as you, the S330, but the microphone will not show in home assistant. Anyone know how to remedy this?