ฝัง
- เผยแพร่เมื่อ 11 พ.ค. 2023
- Learn how to create a python voice assistant that can prompt Bard Ai. This program is free to run. I explain thoroughly how to setup your machine for the python project. The code is explained line by line to help even beginners get a deep understanding of how all of the code works. Learn how to access Google Bard through your Python code. Subscribe for my videos on Python for beginners and AI.
My code for this project: github.com/ai-...
VoiGPT: voigpt.com
Official Ai Austin Discord: / discord
Help me make more videos:
www.buymeacoff...
#programming #python #ai - วิทยาศาสตร์และเทคโนโลยี
![CAMPปลิ้น | EP.79[2/2] มานั่งฮ้านกินข้าวนำกันเด้ออ้ายยย](http://i.ytimg.com/vi/pHIS_Sdy8P8/mqdefault.jpg)
hey I can't make it work im on windows and I got this message
TypeError: Chatbot.__init__() missing 1 required positional argument: 'secure_1psidts'
Bard made a change where you need the token we showed in the video and this token. You can find the token in the same location as the secure 1psid.
Check the Bard API github from the account 'acheong08' for this and any other changes to the API. It will explain how to pass the new token in your code.
@@Ai_Austin thank you boss ‼️
I'm not sure if the reverse-engineered library for Bard changed after you created this guide. I updated code to add a new var and update the call to the chatbot it requires additionally the Secure-1PSIDTS token
tokents = "insert token here"
chatbot = Chatbot(token,tokents)
This seems to work now. New to python and AI so sharing my solutions to anyone else following your guide
Dude keep going this channel is cool!
Thanks for sharing this experiment. Your solution is sending audio from microphone continuously to OpenAI Whisper for hot word detection. This has two issues (1) Everything you say in the room 24x7 is being sent to OpenAI, which means privacy is compromised (2) Whisper API costs money; a year of sending audio 24x7 to Whisper will generate a significant bill.
Hot word detection should be done locally. However current local detection, even with Siri, Alexa and Google Home devices, does not work reliably causing annoyance. This annoyance will be even more if the hot word has to be used in a conversation. There has to be a better way to do this. Solving the hot word problem can upgrade your solution from a science fair demo to a practical product. Keep going at it.
Hey Parul that's actually not true. I'm not using the whisper API. Im using whisper locally. No data gets sent to OpenAI from this program. The only data saved by anyone is the prompt being send to Bard.
@@Ai_Austin I must have missed that. Now it is totally awesome! I need to do it as well. Thank you!
whisper runs locally
Now I just want to buy a Raspberry Pie and try all that to have my own little assistant!
Impressive video, thank you
Exception: Enter correct __Secure_1PSID and __Secure_1PSIDTS value. __Secure_1PSID value must end with a single dot.
I've removed the dot and added the dot. Not sure how I fix it.
you need to have two tokens, the _Secure_1PSID and _Secure_1PSIDTS
Well! Unfortunately i'm not in USA but fortunately VPN exist. 😅
hi austin, can you make a video on how to make a ai (bard or chat gpt doesnt matter) voice assistant for android which can turn on or and off iot devices and answers back to user like any other assistants when you call by name
i would love too to have it on my phone instead
Tasker app. can be configured for that. there are openai plugins in there. I've done it last month
@@JoeTinnySpace yeah but how. Can you explain it to me
I keep on getting the error
"Error transcribing audio: module 'ffmpeg' has no attribute 'Error' "
I've tried uninstalling and reinstalling ffmpeg and ffmpeg-python but it didn't solve the problem. Please help
excelentes videos, me sumo a todos estos proyectos que lanzas... una consulta como hago para poder tener un avatar como el tuyo, preferiría poder elegir una foto y como sincronizo el avatar con la voz? se que existen servicios así como DD pero como te moves en el mundo underground de las IA me imagino que tenes una forma de hacerlo independientemente, sin limitaciones (free or open source xD)
Can you by chance have the response include code to return the voice from elevenlabs? :D
Great video dude, for real, I love your content tbh. I've tried both this one and the last GPT and Bing one but both get me the same ffmpeg issue and i really don't know how to do, i'm kinda new to python.
Terminal says this:
Error transcribing audio: module 'ffmpeg' has no attribute 'Error'
Impressive! i got a challange show us how to create AI assistant with virtuall avatar also it would be cool if you can change name of assistant and have option to speeks in deffrent languages
you can change the name at line 60, where its looking for a specific word, u can change that from bard/google to any other thing. and this is an unofficial wrapper for bard so you cant select languages rn. Hope i could help 😊
I followed everything step by step, I am running this on windows and quite new to python but the error I run into is Error transcribing audio: [WinError 2] The system cannot find the file specified. I checked the folder and the audio file actually gets generated and even listened to it. I wonder why this is happening
I have recently created a virtual AI Science Tutor for kids. Does anyone know where I might be able find an open source interactive avatar like this one to provide a face to my virtual assistant?
Say "Google" to wake me up.
Wake word detected. Please speak your prompt to Bard.
Sending to Bard: can you please name five fruits found in India
Error occurred: 'NoneType' object is not subscriptable
found the solution??
Why didn't you use chromadb for text colourization for Windows?
Nice boss
i am getting this error - OSError: libespeak.so.1: cannot open shared object file: No such file or directory
You need to make a GUI for this!
Are you using wav2lip for your character? If not, what are you using?
how can you do this for a mobile device
Is there a way i can make it hear my voice faster and more effiecently so it ehars me nearly 100% of the time im on windows.
Yes. The speech recognition listen function is by far not the best method as far as reliability works. I'm currently researching and coding out better ways for audio detection.
This is by far not the last voice app video on this channel. Lots of features I want to add and will make videos discussing that in the future.
@@Ai_Austin Thanks I look forward to seeing more videos in the future amazing channel.
So if I understand well, using this method the wake up word can be anything? If so, which Raspberry Pi model you think would be enough to run all that? Knowing that I'll need to plug a microphone, a speaker, and eventually a wifi module on that...
Also, since bard isn't available in my country yet, do I need to constantly run the program through a VPN, or would the VPN only be required during the initial setup?
Yes you can change the wake word to anything. VPN will be required to run always until they release in your country.
As far as raspberry pi, you will want to research how the OpenAI whisper models run on different pi models. One of the newer faster ones will run the code as is with ease. A slower one you may need to use the tiny model for the prompt to make it run faster.
Can you please make a tutorial about implementing bard with Siri in iOS ?
a dumb doubt brother.. will this program only wake up on saying google? cant we set another keyword?
Weird he didn’t try to implement bard too, I mean it’s pretty heavy on my intel mag lol so I get it
Can you do The Reverse engineered API for Quora's Poe? I find Claude to be more conversational.
Nvm I managed to do it
will this code still work as now it is GEMINI
Need to add your system information and software versions, I tried to recreate what you have done but it's not compatible with the newer versions of Python (3.12) I had to search for hours in forums to find this out. and Then I downgraded to 3.11 to get it working.
2:20 😉
please update tha repo so we can have gpt3.5, bing and bard in one program
Bard Ai is publicly?
we got bard in Australia i think almost everyone has it
Yep they gave access to a lot more countries the day this was uploaded.
The AI generated avatar is amazing but the voice uses a tone that looks like he is fighting, I suggest using a more relaxed tone for the voice over .
Sounds to me like he had his coffee. This voice is working perfect for my use. Thanks for your feedback.
and also please reduce the size of the avatar when showing the code, I know it's really amazing and innovative, but when we look at code it's all what we need to see ;)
@alifellahi also please make your own channel or shhhhhhhhhh
I am having this issue "Exception: SNlM0e value not found in response. Check __Secure-1PSID value." . If anybody knows what to do please tell me .
Hey @Ai Austin, great content. Do you have a discord channel?
Not yet, will have one soon.
most of the time when i talk to it it takes my input in arabic even tho i am speaking english can you help ?
how to make speaker face ?
So this is better than Davinci?
Are you on an M1?
Yes
Hi, i don't find the secure_1psidts in my dev instrument, what can i do?
I have the problem, how did you solve it?
Works but extremely slow and seems not to capture my voice at all sometimes. I've check my microphone input levels, all good. Maybe bard is under heavy load?
Running fast for me today! If you can elminate background noise in your room it works best. Sometimes if it detects other sounds it will think you are still speaking. Leading it to keep recording longer than you actually spoke.
@@Ai_Austin OK, thanks. I do have the window open! And thank you for your content, I really like your practical and no-messing/straight-to-the-point style: it's refreshing and easy to watch unlike many channels. Really great work and helping me hit this wave :)
did not work for me. i modified the code according to a comment. STILL NO LUCK
Can we add Operating system commands in this code ?
Absolutely. You could even use function calling to let AI call the commands intuitively or hard code it into your program!
Austin, igot this error :
response = prompt_bard(prompt_text)
response = chatbot.ask(prompt)
PycharmProjects\myAIassistant\venv\lib\site-packages\Bard.py", line 83, in ask
return self.loop.run_until_complete(self.async_chatbot.ask(message))
combine all
Hey great video, I was just wondering why do i keep getting this error? Error transcribing audio: [WinError 2] The system cannot find the file specified. I tried to proofread the code and everything but I can't seem to solve the issue, thank you for your time!
Your getting that error because windows is very crappy for development. Everything code related requires some hack arounds to get it functioning on windows.
It will work, just a matter of diving into that windows specific error. I'd highly recommend installing linux on a hard drive for development and then deal with making windows work after if you want your programs on your windows machine.
@@Ai_Austin great!thank you for your reply i’ll check in to getting linux too
Sorce code ?
Getting this error before bard can respond to my prompt:
Say "google" to wake me up.
Wake word detected. Please speak your prompt to Bard.
Sending to Bard: What's your name? What is your name?
Traceback (most recent call last):
File "C:\Users\tarun\Desktop\AI voice chatbot\BardVoice\main.py", line 98, in
main()
File "C:\Users\tarun\Desktop\AI voice chatbot\BardVoice\main.py", line 88, in main
response = prompt_bard(prompt_text)
File "C:\Users\tarun\Desktop\AI voice chatbot\BardVoice\main.py", line 29, in prompt_bard
response = chatbot.ask(prompt)
File "C:\Users\tarun\AppData\Local\Programs\Python\Python310\lib\site-packages\Bard.py", line 132, in ask
"content": json_chat_data[0][0],
TypeError: 'NoneType' object is not subscriptable
Where did you start learning stuff like this? I want to be able to build stuff like this without tutorials
Start with reading the free book "How to Automate the Boring Stuff with Python" work on it for 1-2 hours at a time. That will give you all of the basics in just a couple weeks.
@@Ai_Austin yea i know about this book, but never read. ig i never needed to as i already did all the projects in the book by myself 😆
The import bard doesn't work. I've followed your tutorial and have installed bard, but it does not import it in my vs code.
Fixed this. I had to lower my python to 3.10
Hi, I recently made BardReadAloud chrome extension to automatically read aloud bard response. Have a try!
i need it for windows broooo pleaseee!!!!
ModuleNotFoundError: No module named 'Bard'
Means you didn't install Bard for the Python versions you are running.
@@Ai_Austin ok thank i already fix it
Exeption has occurred: AttributeError X AttributeError: 'NoneType' object has no attribute 'group' at line 12 chatbot = chatbot(token) (i'm not in usa) compliments for your video austin, any idea how to fix it?
Are you using a vpn connected to a USA or UK server?
@@Ai_Austin to usa
And Bard is working for you in the browser with the VPN?
@@Ai_Austin yes
Has to be an issue with the code or your installation of the Bard library.
⭐⭐⭐⭐🌟
Bro how do I talk to you (while remaining anonymous) about your bots and some help😢 with another broke boy
I do not offer private consultation for free. You are welcome to ask any questions you have in my free public discord.
this tutorial is so confusing because im on windows, please make a bard assistant video for windows.
Programming in general is a lot less fun on windows. I would recommend just not using windows, that's what I will continue doing.
Is this working in india!
from Bard import Chatbot
ModuleNotFoundError: No module named 'Bard'
Is it installed for the python version you are running ?
@@Ai_AustinPython 3.10.2, solved installing all packages manually but now is not recognizing my voice just looping
"Say "google" to wake me up.
No wake word found. Try again."
Its saving the audio correctly but not working for any reason
Has to be a microphone issue or you are speaking before the program started listening.
@@MacacoBellaco how did you manually installed the packeages? I mean python packages or Bard api 's packages?
how did you get unlimited A.I voice from elevenlabs ?
Give them unlimited money 💰
@@Ai_Austin lol
please vosk and how to make speake face ?
😏
Gotta stop ghosting us like that for three months…weeks sorry
hello I am having some trouble here with an error code I was wondering if I need something else to actually make it work " File "C:\Users\Elish\OneDrive\Documents\Bard\main.py", line 12, in
chatbot = Chatbot(token)
File "C:\Users\Elish\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.10_qbz5n2kfra8p0\LocalCache\local-packages\Python310\site-packages\Bard.py", line 63, in __init__
self.async_chatbot = self.loop.run_until_complete(
File "C:\Program Files\WindowsApps\PythonSoftwareFoundation.Python.3.10_3.10.3056.0_x64__qbz5n2kfra8p0\lib\asyncio\base_events.py", line 649, in run_until_complete
return future.result()
File "C:\Users\Elish\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.10_qbz5n2kfra8p0\LocalCache\local-packages\Python310\site-packages\Bard.py", line 143, in create
instance.SNlM0e = await instance.__get_snlm0e()
File "C:\Users\Elish\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.10_qbz5n2kfra8p0\LocalCache\local-packages\Python310\site-packages\Bard.py", line 188, in __get_snlm0e
raise Exception(
Exception: __Secure-1PSID value must end with a single dot. Enter correct __Secure-1PSID value."
I am on windows btw
You seemed to have either typed your access token for bard wrong or didnt add it in at all. Watch back the section of the video for getting your bard access token setup.
@@Ai_Austin oh thank you!
Awesome! How about using ElevenLabs API access for the voice next incl. using different (custom) voices? Didn't watch the full video yet, maybe my question is outdated
I personally wouldnt want to pay 22 a month for a limited voice assistant so am not planning on that.
Try including bark
@@Ai_Austin How about Tortoise TTS?
Say "google" to wake me up.
/Library/Frameworks/Python.framework/Versions/3.11/lib/python3.11/site-packages/whisper/transcribe.py:114: UserWarning: FP16 is not supported on CPU; using FP32 instead
warnings.warn("FP16 is not supported on CPU; using FP32 instead")
Error transcribing audio: Could not load sound with filename, although URL was good... file:///Users/zaursamadov/Desktop/KGB_45/wake_detected.mp3
I'm keep seeing this error. do you know how to fix it sir?
You're trying to transcribe the wake_detected.mp3 file. That should be the beep notification sound file.
The wake word detection code is supposed to transcribe the wake_detect.wav file.