Build Your Own AI Assistant Part 1 - Creating the Assistant

[@kevinmcaleer28]

What skill shall we work on next, Weather? Home Automation? Calendar? Time (as in 'what time is it'), Random Facts?

[@SvenzAB]

Something that Will use elastic research?

[@kevinmcaleer28]

@@SvenzAB yes - I’ll look at that next Alfio!

[@xx8031]

Some time in the future, I'll be working on including a chat or UI with Google assistant or Cortana and giving thoes entities an image.

[@rebelion1]

what are you using to start this project in vscode? I dont get that part

[@kevinmcaleer28]

@@rebelion1 Do you mean how do I run the code? I use the Green play button at the top right of the screen

[@ayushojha6727]

I think I finally found a way to do my final year project. Please complete this series. You seriously are a lifesaver. Edit 1: You have everything I was thinking to include in my project. Edit 2: There are more videos on this series. I am so glad I found your KZbin Channel. As I continue the video I am more and more grateful towards you.

[@monikachivate6287]

Hey I also want to do this project... Can we work together?

[@tevanize]

Mind sharing a little on ur project?

[@LexifromZargon]

yo howd it go? wanting to do the same for my final project

[@ayushojha6727]

@@LexifromZargon Not very well guys. I ended up making a simple chatbot. Almost screwed my GPA.

[@brightmurima665]

@@ayushojha6727 oh sorry to hear that, do you mind sharing some of the challenges that you faced?

[@jacktringoli3299]

How would you go about improving the voice? Is that a type of module or software to get a better sounding voice? Or is there some kind of way to have the ai generate the voice entirely on its own like siri?

[@kevinmcaleer28]

Yes - there are options in the voice recognition library to use on device recognition (I guess it’s just not as good as Google) but works if you don’t want to send off data to the cloud.

[@RATCHET0697]

I think there'd be software for it to improve or make it sound how you want like a celebrity, character or even an accent

[@akulgoel9259]

please make nlp integration with this, i can see this become a powerful tool someday

[@kevinschilling6813]

Hey Kevin.. I'm coming a little late to the show so to speak.. I was following along great right up to the Demo Time.. I thought you were going to VS on your computer not through the Raspberry Pi environment. Which makes things a tad different.. I will have to come back to the Demo Time point after my mic, speakers and micro HDMI to HDMI cable comes in the mail.. I do like how detailed you are with things..

[@AlleAnneWand]

where to get the head?^^ (frankenstein)

[@Hybrid.Robotics]

Yes, I would also really like to see more videos on microcontrollers and robotics. More Micropython would also be awesome!

[@TheSteveMorganShow]

can i do this on my pi 1 model B+

[@MagDag_]

@Kevin McAleer Thank you for tutorial! I have a Nvidia Jetson nano will try to use it.

[@flankz2950]

You are such a legend! was using this for my prject

[@willmaticio]

love this video, I'm still in the process of finishing as I'm working to create something similar. as far as the speaker and mic, have you thought about using a bluetooth headset or earpods as the form of giving instruction and receiving output

[@rodriguezmartin496]

Hey, It looks amazing my friend, good job. Can you do one with a raspbeery pi pico?

[@spencerhammond2870]

Still having trouble with the import speech_recognition part. It says there's no module named that, but I installed it. Any advice?

[@kevinmcaleer28]

I can help you better over in the small robots Facebook group - are you a member of that?

[@cesargodinez1614]

try: phrase = self.r.recognize_google(audio,language="en-US", pfilter=0, show_all=False) sentence = "Got it, you said" + phrase return phrase self.engine.say(sentence) self.engine.runAndWait() print("You said: "+phrase) except sr.UnknownValueError as e: print("Sorry, didnt catch that",e) self.engine.say("Sorry, didnt catch that") self.engine.runAndWait() ----------------------------------------------------------------------------------------------------------------------------------------------- This happened to work for me without bringing up the e is not defined and also returning within the try block

[@connorhatchs]

you are a life saver. i've been looking for a solution for way too long.

[@cyphertech7856]

@Kevin McAleer can I programming just the software along without including the raspberry

[@Mport-UK]

Hello mate, I have only just came a cross you tbh and Just stated watching, but for what I can see you are making very high quality videos! Truly thank you

[@ImpactBST]

When I Try To Run Mine It Runs Then Just Ends It Straight Away Doesnt Give Me Time To Say Anything 😥

[@ImageFoxStudios]

Can i make an ai that has my mother's voice and speaks Sinhalese ( Sinhalese is a lanuage in srilanka and can you cket is it available in google voices )❤️

[@mrzfpv7871]

your tutorial helps me a lot as an electrical engineer developing delivery drones

[@kevinmcaleer28]

That’s great

[@compubyte2010]

yeah Microcontrollers. especially the PICO.. I really feel the PICO W blows away the 8bit Ardunio, and can elevate those things built with it. to another level! IMO

[@Mport-UK]

I just keep getting this like this Exception has occurred: NameError name 'e' is not defined NameError: name 'self' is not defined class AI(): NameError: name 'e' is not defined How can self not be defined? :'( I have made sure indents are in the right place but still get the something :'(

[@eDRoaCH]

Absolutely loving these. I am not fully sure because of the nature of video but I think your crash problem is that if the Try in listen() fails you have nothing to return.

[@kevinmcaleer28]

Thanks! Yes, when nothing is returned it raises an ErrorUnknownValue - I'll need to catch this in the exception block and get it to say 'sorry I didn't catch that'

[@clarencenikel_6420]

Hi Kevin, I was wondering if I can create something using this program to replace or maybe fix something that is broken. But I’m more into replacing a broken structure.

[@kevinmcaleer28]

Clarence - do you have anymore information about what you’re looking to replace or fix?

[@Jennn]

4:33 This is such a clever and charming way of illustrating a point. I love it! I had originally just listened to this video at work in my ear phones, finally had the time to come back and actually watch it and glad I did. 🙂Thank You for this, looking forward to watching / listening to more.

[@kevinmcaleer28]

Thanks!!

[@ManjuChoudhary-ww1wn]

Instead of using the USB microphone can I use a cardioid condenser microphone with audio amplifier and an extra wireless mic, so I can give commands

[@patloeber]

Great video!

[@kevinmcaleer28]

Thanks! Happy to take your feedback on any code mistakes or stylistic errors

[@surendrakumarsahoo9878]

Hi, I am sahu from Bangalore India. I want to join, please let me know about it. Thank you 🙏❤🙏

[@CraigTinson2015]

Excellent video... and thank you... and this seems closer to what I've been researching for the last few weeks... but I've already hit a hiccup. I've seen you've restarted the project as of a couple of days ago so I figured this would be the most up-to-date basis I could work from. However, using the your requirements.txt all the pyobjc requirements fail (and there is a lot of them).. so after some research I found that these are for mac's only?? before I go any further into this tutorial is that going to become an issue considering I'm doing this on a Raspberry PI 4B? or will this be explained? I've done your new vosk model changes and that all worked well.

[@kevinmcaleer28]

Hi Craig - you can remove all the objc references. They get installed by the speechrecognition module as it’s a wrapper library that pics the correct modules for your specific hardware. If I remember I’ll remove those and test it on my RPi 400

[@albertomejia1570]

I did everything you said but when I test it like you do at 50:58 it doesn't respond or it doesn't pick up my mic audio, maybe? I know it works because I tested my mic on Google voice search. Maybe, I'm doing something wrong?

[@kevinmcaleer28]

Are you using a usb microphone - can you see that with arecord - raspberrypi.stackexchange.com/questions/105280/identify-which-microphone-is-on-which-audio-card-using-arecord

[@Hybrid.Robotics]

I have an 8Gb Raspberry Pi 4 with an Adafruit Braincraft Hat that I am installing all of this on. It boots from a 128Gb USB drive. :)

[@essazouga1390]

Can you integrate this with computer vision, labeling and pose estimations?

[@kevinmcaleer28]

Absolutely- I’ve done quite a few videos on using OpenCV to detect objects, labelling and faces etc kzbin.info/www/bejne/oXvTfYB4prOWf7M

[@alexbrockman731]

I think being able to turn on vehicle in the morning would be a good skill

[@kevinmcaleer28]

I have a renault zoe, and there is an app that allows you to do that (warm it up). It uses an API, so all I'd have to do is push a POST to the API and hey-presto I can warm my car using python or node-red...

[@iotimmiey]

would this work on android 11 or ubuntu touch? i have a pixel 2xl that would be great for this

[@karnivijay7112]

Hii sir

[@dionisgedeshi7834]

can you run this project on your computer, without the need for a raspberry device? I appreciate the videos, keep up the great work

[@kevinmcaleer28]

Yes - this works on Linux, Mac and Windows

[@dionisgedeshi7834]

@@kevinmcaleer28 Thanks for the help, I will definitely stick around this tutorial. You are such a good creator, that replies to all the comments

[@atelier_mongame4119]

Mlk, se pá que o canal foi hackeado

[@kevinmcaleer28]

Why do you think that?

[@MsShaiksadiq]

hi , how to contact you . I need help in setting this up for my university project

[@makafuiaddiah1238]

Your voice is so cool and calming

[@the1andonlycw]

Thanks so much for this! I'm starting to learn python for this very reason.

[@compubyte2010]

you mentioned FLASK for Web UI .. do you recommend this over PHEW?? ( Hoping to find a way to connect a PICO to an external drive! .. so I can store info and stuff.. etc.. )

[@kevinmcaleer28]

You can connect the Pico to an SD Card, there are quite a few addons out there for this.

[@Catamox]

I know this is an old video but this is absolutely amazing

[@kevinmcaleer28]

Thanks!!

[@Stegeno]

Hi this Ai can be a friend or only an assistant

[@Clanslayer-kj7rk]

I think it would be a lot better if we use "Sapi5"? Wouldn't it be?

[@ItsJayBirds]

Which Rasberry Pi did you recommend, 4 or 400? Also how about weather and calendar for next skill?

[@kevinmcaleer28]

400 is great if you want it all in one box, the 4 is great if you want to run it headless (no monitor, keyboard etc)… I’ll definitely look at weather next as a skill then calendar

[@ItsJayBirds]

@@kevinmcaleer28 Thank you for replying quickly!

[@tradingwithbots4707]

Your are the best, you're probably a professor aswell.

[@fw1engineer]

I know this tutorial is 2 years old but i was walking through it and at the simple test "tell me a joke" i have the code exactly as you do but it doesn't match: Microsoft David Desktop - English (United States) Listening new todo list created Say Something Got it You said tell me a joke command was: tell me a joke Say Something Got it You said goodbye command was: goodbye and if statement is: if command == "tell me a joke": joke() but id doesn't match??

[@kevinmcaleer28]

Do you have a GitHub repo with your code in so I can see?

[@karthikchowdary1660]

Can i make the same AI in laptop without using raspberry pi as my laptop has faster chipset than raspberry pi?

[@kevinmcaleer28]

Yes - during later videos I use my Mac to host the AI. You can also use A PC running Windows or Linux too

[@karthikchowdary1660]

@@kevinmcaleer28 dear kevin , i wanna get your guidance on some project I’ve planning to do and it’s really important for me.. and I believe you’re the right person for the job … how can i contact you?

[@kevinmcaleer28]

@@karthikchowdary1660 The best way is via the Small Robots group on facebook - facebook.com/groups/smallrobots

[@karthikchowdary1660]

@@kevinmcaleer28 can you check there once i pinged you jus now

[@boomatics1365]

Salute to you u made a 1 hour video and have only 3.3 k

[@AkshajNallamothu]

What website did you use in demo time

[@nabojyoti9189]

Chatgpt directed me here

[@kranstopher]

I'm running python 3.8 will the AI still work, or do I have to mod the code the work?

[@kevinmcaleer28]

Python 3.8 should be fine, no mods required

[@4clockfarms]

How do I add a face to this AI on my computer? Can it be connected to an animation like Max Headroom?

[@kevinmcaleer28]

If you're good at animation I'm pretty sure you could combine to the two together

[@xx8031]

Thank You Kevin. What adjustments might I need to make to the code to run python ai on a PC laptop?

[@kevinmcaleer28]

I’m running that code on a laptop so give it a try and let me know how it goes

[@xx8031]

Thank You Kevin. I'll let You kow how things go.

[@xx8031]

@@kevinmcaleer28 So, Kevin are you saying that I can just download the code and run it? I've unzipped it I'll look into running it.

[@xx8031]

@@kevinmcaleer28 Heykevin, I've installed VS; looks like I can clone it. IDK where to go from there. I'll investigate it.

[@ICEDANY36CHANNEL]

How much RAM does raspberry need for AI? Because I don't know whether to take the 2GB or the 4GB one

[@kevinmcaleer28]

It’s depends on the size of the model you’re using or training. I’d get the largest ram you can afford, if you can

[@cpc123ify]

Could an AI be taught math science Small Time programming and so on

[@kevinmcaleer28]

AIs aren’t conscious so can’t be ‘taught’ anything, - the algorithm can be trained to spot patterns but that’s not the same as creative thought. AI is really just a function - you put stuff in and it processes it to produce an output. You couldn’t give it a problem and it produce a program as an output.

[@mohamedrusthum8924]

can we make it so it doesnt need a wake word and is always listening and noting down wat i am saying is this feasible?

[@kevinmcaleer28]

You can! There will be a pause between listening and transcribing, so anything said between those steps will be missed

[@satyakitalukder7024]

you are a legend man! the vdos are so unique and procedural! like them very much

[@kevinmcaleer28]

Thank you

[@husshuss4776]

THIS IS SO COOL HAAHA

[@Hybrid.Robotics]

Kevin, move the say() up near the top. Then you can just use self.say(sentence) instead of retyping those two lines over and over.

[@kevinmcaleer28]

Thanks for the code hint! I have a mantra when it comes to programming - make it work, make it right, make it fast. In this case I only made it work!

[@rajeevranjanpratapsingh1840]

i want both wake word ai assistant and non-wake word ai assistant

[@kevinmcaleer28]

Yes you can do that - the AI will constantly look for words so just decide one to be the wake word (or just skip that part completely)

[@dtteamofficial]

fun fact: I coded my Alexa skill "Dani Translator" in python.

[@cullenbuteau2143]

Can you make a skill were we can play music for free

[@kevinmcaleer28]

I'm sure you can, but not for copyrighted music, for that you'd need a deal with the music distribution companies

[@empelandogropht1329]

merci beaucoup a toi :))

[@realxd553]

I'm getting some error

[@Alcazar_official]

Does making robots need math?

[@kevinmcaleer28]

It certainly doesn’t hurt when your working out inverse kinematics…. But you can build robots without math

[@marknantwi5662]

I wanted to create a robot 1.Should I use raspberry pi 4 2.can I contact you privately please I want to express my idea to you

[@SkySesshomaru]

amazing!!!!!!!!!!!!!!!!

[@saladbarchatswood3212]

Any lists of hardware we needed?

[@kevinmcaleer28]

Just a computer, microphone and speaker. Any Raspberry Pi, Linux computer PC or Mac should be fine

[@loginnow7110]

The voice of assistant is so bad can we change it?

[@kevinmcaleer28]

Yes - there are other voices available - its probably more the quality of the speakers and my microphone; its quite audible in real life

[@compubyte2010]

Okay I have to ask, cuz my Ignorance on this stuff.. LOL .. But.. you said to do the AI, you need as fast as you can get RPi 4 etc.. But the AI of Google Home's electronics aren't close to a Pi4 ? but it seems to work fast enough?? Or am I missing something? LIke there AI is on a server somewhere?? which I could use an old laptop for a AI instead of a RPi? I can install the Linux on the laptop.. ( Install RPi on an old Laptop.. ran very fast :) .. IF that will work? I just want my AI ( head. ) to talk to me like google but much better. more personal for ME.. to be more technical and help me with electronics etc. I say all this , cuz as you know getting a Rpi P4 these days is almost impossible Unless you want to spend $200 for one!! I mean I have some.. but..trying to figure out if I can use some other electronic pc's I have. NUC laying around.. etc..

[@kevinmcaleer28]

You can use what you have. Since I made this video Google have stopped the free voice to text service so now you’ll need a computer that can run VOSK which does it on chip, it has different sized models depending on how much RAM you have available. An old laptop should work fine with Linux on

[@compubyte2010]

@@kevinmcaleer28 the free text to voice? well .. I'm not sure thats what I used at google? I have one of those google voice boards that you can progam and use as your own google home .. some HAT for the Rpi. .forgot what it's called.. i'm sure your familiar with it? .. I was going to use that.. but I do have about a dozen old laptops I could use.. old P4/ i3 or i5's?? forget LOL

[@tubbydrunk6343]

Hi everyone, I tried to redoing it on my own on a virtual machine with Ubuntu on my win10 pc but when i run it , it gives me an attribute error: 'AI' object has no attribute 'listen'. Has someone had my same problem? I tried even on windows, enviroment where i feel a bit more comfortable but i still recive the same error.

[@kevinmcaleer28]

Can you share your code? It’s likely to be either a missing dependency or missing import statement

[@tubbydrunk6343]

@@kevinmcaleer28 import pyjokes from ai import AI Alf = AI() def joke(): funny=pyjokes.get_joke() print(funny) Alf.say(funny) command="" while True and command !="goodbye": command=Alf.listen() print("command was:", command) if command =="Tell me a joke": joke() Alf.say("Goodbye,I'm going to sleep now") Here is the code of the second file.

[@tubbydrunk6343]

import pyttsx3 import speech_recognition as sr class AI(): _name="" _skill =[] def __init__(self, name=None): self.engine =pyttsx3.init() self.r = sr.Recognizer() self.m =sr.Microphone() if name is not None: self.__name =name print("Listening") with self.m as source: self.r.adjust_for_ambient_noise(source) @property def name(self): return self.__name @name.setter def name(self, value): sentence="Hello, my name is"+self.__name self.__name = value self.engine.say(sentence) self.engine.runAndWait() def say(self, sentence): self.engine.say(sentence) self.engine.runAndWait() def listen(self): print("Say Something") with self.m as source: audio=self.r.listen(source) print("got it") try: phrase = self.r.recognize_google(audio, show_all=True, languege="en_Us") sentence = "Got it, you said" + phrase self.engine.say(sentence) self.engine.runAndWait() except any as error: print("Sorry, didn't catch that", any) self.enginge.say("Sorry didn't catch that") self.engine.runAndWait() print("You said", phrase) return phrase And here is the first one

[@EmosGambler]

Starting now! Thanks for such content. Hopefully I will manage!

[@photomaster1]

This looks more like an expert system no AI system.

[@dan_rad]

I really like your presentation, non-standard shaped video bubble and nicely integrated with the presentation. How did you create this?

[@kevinmcaleer28]

I created a custom overlay and use Ecamm Live to present the video, and a Streamdeck to switch scenes & cameras. Thanks Daniel

[@shaymanorigins3412]

if you say nothing he runs into the error loop, with e as error, but this grants you an error breaking the loop, how to fix that, and i also noticed by saying nothing the loop also breaks thanks to the same error, and the command.lower() will break it too, any idea how to fix that? i aint the best at python, love the tutorial btw

[@kevinmcaleer28]

As this is the first video in the series I used Google voice services, which was free to use at the time. In a later video I switch to using Vosk which is free and doesn't need Google - the error may be due to that.

[@shaymanorigins3412]

@@kevinmcaleer28 ah cool, appreciate your vids a lot, gonna drop a sub, so following the tutorial we will switch to vosk too?

[@ManojAP]

How would I handle the e error (except)

[@kevinmcaleer28]

Hi! the e error exception is a generic error so its difficult to troubleshoot via KZbin. In later episodes of this series we switch to using Vosk for voice recognition as Google has stopped the free tier for voice services. It may be that this is causing the issue. See the other episodes in the series to fix this

[@AMIRSHOOTER]

NICE ! Keep going

[@totalmaniac7717]

Is this compatible with python 3.11? Because every time I run the program it says "Process finished with exit code 0"

[@connorhatchs]

that means the code executed properly.

[@xx8031]

Thank You Kevin. What adjustments might I need to make to

[@TimFabianFiko]

Only me who have a problem with "except e as error: NameError: name 'e' is not defined"

[@kevinmcaleer28]

Does it say which line number?

[@connorhatchs]

im having this same issue. did you ever figure it out?

[@TimFabianFiko]

@@connorhatchs No, i just gave up. Sorry

[@dracone4337]

Any way to do this without a raspberry pi

[@kevinmcaleer28]

Yes - this code works on macOS, Windows and Linux computers

[@lostinkansasonasunnyday305]

All that is needed is a computer and Python 3.

[@aedaldaniel]

AI from 1980s

[@kevinmcaleer28]

It’s aimed at people who haven’t created one before.

[@wicklif1]

Wow this is what have been looking for

[@cpc123ify]

Will this work with Arduino boreds ?

[@kevinmcaleer28]

Unfortunately not - the Arduino is an 8bit board with barely any RAM, AI needs a lot of RAM and processing capability (32bit chip or better typically). The Raspberry Pi Zero is cheaper than an Arduino, so theres that.

[@cpc123ify]

@@kevinmcaleer28 ok thank you very much for the info I can't wait to my AI

[@kevinmcaleer28]

@@cpc123ify let us know how you get on

[@dailychinese1396]

thanks amazing video! I loved it !

[@cullenbuteau2143]

can you make custom skills??

[@kevinmcaleer28]

Absolutely - what kind of thing do you think I should add?

[@cullenbuteau2143]

@@kevinmcaleer28 Could you make a skill were it help looks for coding scripts and teaches you how to code??

[@RAJ-nt7xr]

Hi , will it work on windows ? If yes then how to install the pre-requisites in windows ?

[@kevinmcaleer28]

Yes it will work on Windows too. The prerequisites are the same - the Python Speech Recognition pip library is a wrapper and will pull the relevant platform specific files for you.

[@RAJ-nt7xr]

@@kevinmcaleer28 thankyou. !! i will try to work on it .....you got a new sub. :D ❤

[@LUXSON_Tech_Demos]

how do I choose what microphone to use?

[@kevinmcaleer28]

Hi Leo - What Microphone do you have? I bought a really cheap one from Amazon, but honestly you could just use the handsfree headset one you get with your phone. The voice recognition is pretty reliable (especially if youre using google's voice service like I do in the demo)

[@LUXSON_Tech_Demos]

@@kevinmcaleer28 Samson Go mic. It’s card 3 and device 0

[@LUXSON_Tech_Demos]

It Definitely works as Iv tested the arecord command however my guess is that python doesn’t know which card to use for the microphone

[@LUXSON_Tech_Demos]

Do you have any ideas I’m dying to do this project

[@kevinmcaleer28]

@@LUXSON_Tech_Demos I understand your issue now, let me have a look at this and test a couple of things

[@cyphertech7856]

@Kevin McAleer can you help me i got this Got it! Now to recognize it... Uh oh! Couldn't request results from Google Speech Recognition service; recognition connection failed: [Errno 11001] getaddrinfo failed

[@kevinmcaleer28]

Google have recently stopped this as a free service. I've recently changed to using Vosk as an alternative - you can find that video here: kzbin.info/www/bejne/fWTHZ5R8ftSbhtE

[@cyphertech7856]

Then how do I fix it from the beginning Can I chat you privately please 🤲