Thanks! How does this compare to the Conversational Speaker project? It seems that this method doesn't require one to pay for Azure Cognitive Speech Services, which is great.
That would be a completely different project. If you're interested in pretraining a model from scratch please reach out to me through my discord (its in the github). I can show you how to scrape 570gb of text similar to what openai uses for their text corpora. Wouldn't use openai API and also not viable on a raspberry pi due to the insane computational resources required.
im not too familiar with the pico, thats something you could research. if you have speakers and a decent microphone there is a decent chance it would work with your humanoid project. thanks for commenting!
Superb! Thanks for this standalone personal assistant that requires no PC / Laptop once everything is loaded to Raspberry Pi. Few questions--1) Can Raspberry Pi 3 be used as availability of RP4 is very bad. 2) Can you please add a camera to recognise each face and greet the person by his / her name accordingly instead of a common name like "Elliot" which is applicable to the user only?
Thanks! To answer your questions, any raspberry pi should work. You may receive slightly different errors but likely the same ones. Haven't tested on every model so not 100% sure. You could implement your own computer vision algo that would recognize your face. That becomes a bigger problem because raspberry pi's aren't powerful and would either lag or crash when given difficult tasks like computer vision scripts. You could try to implement this on a desktop PC or a laptop since they can better handle it. In terms of the name, you can actually change the name variable inside the python script (located in the github repo in the video description). I might consider trying out the face recognition though. Good idea!
That is possible. What you would look into is making the script run on boot, so that you don't have to cd into the directory. Haven't looked into that on raspiberry pi yet. Try searching for "run python script on boot raspberry pi"
You should be able to ssh into the pi assuming the ssh service is setup. It's well documented you can likely figure it out in your own from here. Ssh let's you control the pi from a remote device
Well, I'm thinking to make a talking robot, and use raspberry pi to access chatgpt and gives the output...But I'm unclear about these stuffs...Tell me how can I connect with you on Social Midea?@@elliotarledge
Thanks again. After a lot of wrong entries made by me by mistake and then watching your video multiple times I could complete the project successfully. The only problem is that the audio output is too robotic. Hope you shall update the project soon. Finally, I got one error that asked to install flac. After doing that everything worked ok.
Great job! This project isn't perfectly optimized for everyone yet. I also run this on a raspberry pi I got many years back so there are a few things mine has/doesn't have compared to more recent ones. As for the voice, those were the only devices that came with the OS. If you really wanna upgrade the quality, consider downloading new voices and setting them up properly in the registry.
Hi when I start the code I get the error " line 18, in voice = engine.getProperty('voices')[1] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^ IndexError: list index out of range" could you maybe help me? I'm a little stuck atm :(
change the index [1] to [12]. if that doesn't work try [17] instead. a few other people had this problem too. i setup my raspi OS a few years back so probably forgot something I changed back then. thanks for reaching out!
@@vincentgroen4707 add this line under the one you are getting errors `print(engine.getProperty('voices'))`. you will get a ton of voices printed out in the form of an array. If you arent familiar with those, you should either reach out directly to me on discord, or learn how arrays work then choose the right index according to your preferred language. Try [0] instead of [1]
Super your video but ,, :-) At this step i have this error ==> arecord -f cd -D hw:1 | aplay error --> pi@rpiChapGPT:~ $ arecord -f cd -D hw:1 | aplay arecord: main:830: audio open error: Device or resource busy aplay: read_header:2839: read error
"Step by Step" would include explaining your development environment. What is this screen you're showing us, Is it a terminal, a command shell, a jupyter note book??? I believe you're video is getting lost in the ocean of others ramblings because it is not "step by step" similarly to the other 100,00 videos that make zero sense to newcomers and they stop watching after the first cpl of minutes,. Just because you know what your doing doesn't mean others do lol.
awesome broo ❤️ everything was super simple and easy to do except that api key stuff
Good stuff, Just subscribed, keep it up man
Good job bro
New sub.. nice my dude.
Hi! Looks awesome! I’m new to this so I don’t know much about raspeberry pies so which one did you use?
I have an older version from 5 years ago or so. To run scripts that aren't computationally heavy, I'd recommend the Raspberry Pi 2.
@@elliotarledge thank you!
Hii Sir this project is work on raspberry Pi b model 3
Thanks! How does this compare to the Conversational Speaker project? It seems that this method doesn't require one to pay for Azure Cognitive Speech Services, which is great.
this is chatGPT but purely voice for the utility. enlighten me on the conv speech project you are referring to!
Hii this project is work on raspberry Pi b model 3
now, how do we hook this up to a furby hahaha
Thanks for sharing :)
which version of Pi, you use?
Subbed! 👊
Is it possible to have it running offline? I don’t mind downloading the 570gb library
That would be a completely different project. If you're interested in pretraining a model from scratch please reach out to me through my discord (its in the github). I can show you how to scrape 570gb of text similar to what openai uses for their text corpora. Wouldn't use openai API and also not viable on a raspberry pi due to the insane computational resources required.
Can I do this on a pi pico ? With external SD card if necessary? And insert it into a robot humanoid I have made? Pls reply the possibility
im not too familiar with the pico, thats something you could research. if you have speakers and a decent microphone there is a decent chance it would work with your humanoid project. thanks for commenting!
Superb! Thanks for this standalone personal assistant that requires no PC / Laptop once everything is loaded to Raspberry Pi. Few questions--1) Can Raspberry Pi 3 be used as availability of RP4 is very bad. 2) Can you please add a camera to recognise each face and greet the person by his / her name accordingly instead of a common name like "Elliot" which is applicable to the user only?
Thanks! To answer your questions, any raspberry pi should work. You may receive slightly different errors but likely the same ones. Haven't tested on every model so not 100% sure. You could implement your own computer vision algo that would recognize your face. That becomes a bigger problem because raspberry pi's aren't powerful and would either lag or crash when given difficult tasks like computer vision scripts. You could try to implement this on a desktop PC or a laptop since they can better handle it. In terms of the name, you can actually change the name variable inside the python script (located in the github repo in the video description). I might consider trying out the face recognition though. Good idea!
@@elliotarledge Thank you so much
one giga ram is it will work?
Bro i got a school project can you make it like when we star the Rasspberry pi it that voice assistant start working
That is possible. What you would look into is making the script run on boot, so that you don't have to cd into the directory. Haven't looked into that on raspiberry pi yet. Try searching for "run python script on boot raspberry pi"
Can I have ur watsapp noi want to talk to u??
Yo bro....Should I connect it to my PC to operate it ?...
Can't I upload the code and connect it to the battery for powersupply and do the stuffs...
You should be able to ssh into the pi assuming the ssh service is setup. It's well documented you can likely figure it out in your own from here. Ssh let's you control the pi from a remote device
Well, I'm thinking to make a talking robot, and use raspberry pi to access chatgpt and gives the output...But I'm unclear about these stuffs...Tell me how can I connect with you on Social Midea?@@elliotarledge
Can i do it in Spanish?
Thanks again. After a lot of wrong entries made by me by mistake and then watching your video multiple times I could complete the project successfully. The only problem is that the audio output is too robotic. Hope you shall update the project soon. Finally, I got one error that asked to install flac. After doing that everything worked ok.
Great job! This project isn't perfectly optimized for everyone yet. I also run this on a raspberry pi I got many years back so there are a few things mine has/doesn't have compared to more recent ones. As for the voice, those were the only devices that came with the OS. If you really wanna upgrade the quality, consider downloading new voices and setting them up properly in the registry.
Thanks. There was another small error in the code. "hey pos" should be single quotation mark. Ie 'hey pos' in text.lower():
I am using Raspberry Pi 4 B, 8GB with a 256 GB micro SD card. I think if I load OS Windows 10 then I can opt for better audio like Davinci
hey i have a question. is it possible to change the voice?
yes change the index of the voice property
@@elliotarledgehow do you that?
Hi when I start the code I get the error " line 18, in voice = engine.getProperty('voices')[1] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^ IndexError: list index out of range" could you maybe help me? I'm a little stuck atm :(
change the index [1] to [12]. if that doesn't work try [17] instead. a few other people had this problem too. i setup my raspi OS a few years back so probably forgot something I changed back then. thanks for reaching out!
@@elliotarledge mhh it still gives me the error :( do you know any other ways to fix or trouble shoot this? (+ Thank you for responding!)
@@vincentgroen4707 add this line under the one you are getting errors `print(engine.getProperty('voices'))`. you will get a ton of voices printed out in the form of an array. If you arent familiar with those, you should either reach out directly to me on discord, or learn how arrays work then choose the right index according to your preferred language. Try [0] instead of [1]
@@elliotarledge Thank you! The code is working now :D
From we're can I git a API
Plz answer me
@@farahkhalil6846 Hey! After creating an account on the openai website you can figure out API stuff in your settings on the openai website.
Super your video but ,, :-) At this step i have this error ==> arecord -f cd -D hw:1 | aplay error --> pi@rpiChapGPT:~ $ arecord -f cd -D hw:1 | aplay
arecord: main:830: audio open error: Device or resource busy
aplay: read_header:2839: read error
Just asked gpt-4 the answer, you should do the same. Would be inefficient for me to send instructions. Thats also part of being a programmer :D
Do you think it would be possible to make this speak in Spanish?
Yes just look up different voices. Shouldn't be easy since the OS has a bunch by default
He looks like a Brandon
can i have ur no
I don't publicly give out my phone number but you can add me on discord.
@@elliotarledge okk ur id
I hate the voice, I’d like to make it better
"Step by Step" would include explaining your development environment. What is this screen you're showing us, Is it a terminal, a command shell, a jupyter note book??? I believe you're video is getting lost in the ocean of others ramblings because it is not "step by step" similarly to the other 100,00 videos that make zero sense to newcomers and they stop watching after the first cpl of minutes,. Just because you know what your doing doesn't mean others do lol.
you're right. this is why I started working on a new one. appreciate the honesty :)
ur discord id
Infatoshi Nakamoto#3233
@@elliotarledge accept request from anmol