Skip to main content

JarvisAI is python library to build your own AI virtual assistant with natural language processing.

Project description

Header

PythonTensorFlowPyTorch

Hello, folks!

This project is created only for those who are interested in building a Virtual Assistant. Generally, it took lots of time to write code from scratch to build a Virtual Assistant. So, I have built a Library called "JarvisAI", which gives you easy functionality to build your own Virtual Assistant.

Content-

  1. What is JarvisAI?
  2. Prerequisite
  3. Architecture
  4. Getting Started- How to use it?
  5. What it can do (Features it supports)
  6. Future / Request Features
  7. Contribute
  8. Contact me
  9. Donate
  10. Thank me on-

Premium Plan-

What is our premium plan?

  • AI will be able to understand all your commands. It will answer all your questions apart from below basic intent.

  • It will be able to handle intent- 'others / Unknown Intent'. Free plan doesn't support this.

  • It will be automatically upgraded to use GPT-3 based model in the future. Currently, it uses other advance custom AI models to answer queries.

  • Currently unlimited API calls. Later we might change / limit.

  • Currently, it doesn't remember the previous context of the chat, but soon it will be. We don't store your personal chat information.

Check out our plan: https://jarvisai.in/dashboard

YouTube Tutorial-

Click on the image below to watch the tutorial on YouTube-

Tutorial Latest-

JarvisAI Tutorial 1

Tutorial 1-

JarvisAI Tutorial 1

Tutorial 2-

JarvisAI Tutorial 2

1. What is Jarvis AI?

Jarvis AI is a Python Module that is able to perform tasks like Chatbot, Assistant, etc. It provides base functionality for any assistant application. This JarvisAI is built using Tensorflow, Pytorch, Transformers, and other open-source libraries and frameworks. Well, you can contribute to this project to make it more powerful.

2. Prerequisite

  • Get your Free API key from https://jarvisai.in

  • To use it only Python (> 3.6) is required.

  • To contribute to the project: Python is the only prerequisite for basic scripting, Machine Learning, and Deep Learning knowledge will help this model to do tasks like AI-ML. Read the How to Contribute section of this page.

3. Architecture

The JarvisAI’s architecture is divided into two parts.

  1. User End- It is basically responsible for getting input from the user and after preprocessing input it sends input to JarvisAI’s server. And once the server sends its response back, it produces output on the user screen/system.

  2. Server Side- The server is responsible to handle various kinds of AI-ML, and NLP tasks. It mainly identifies user intent by analyzing user input and interacting with other external APIs and handling user input.

    JarvisAI’s Architecture

4. Getting Started- How to use it?

NOTE: Old version is depreciated use latest version of JarvisAI

4.1. Installation-

  • Install the latest version-

     pip install JarvisAI  
    

Optional Steps (Common Installation Issues)-

  • [Optional Step] If Pyaudio is not working or not installed you might need to install it separately-

    In the case of Mac OSX do the following:

     brew install portaudio  
     pip install pyaudio  
    

In the case of Windows or Linux do the following:

  • Download pyaudio from: lfd.uci.edu/~gohlke/pythonlibs/#pyaudio

  • pip install PyAudio-0.2.11-cp310-cp310-win_amd64.whl

4.2. Code You Need-

You need only this piece of code-

import JarvisAI

def custom_function(*args, **kwargs):
    command = kwargs.get('query')
    entities = kwargs.get('entities')
    print(entities)
    # write your code here to do something with the command
    # perform some tasks # return is optional
    return command + ' Executed'


jarvis = JarvisAI.JarvisAI(input_mechanism='text', output_mechanism='text',
                       google_speech_api_key=None, backend_tts_api='pyttsx3',
                       use_whisper_asr=False, display_logs=False,
                       api_key='99f605ce-5bf9-4e80-93a3-f367df65aa27')
JarvisAI.add_action('custom_function', custom_function)
jarvis.start()	

4.3. What's now?

It will start your AI, it will ask you to give input and accordingly it will produce output.
You can configure input_mechanism and output_mechanism parameter for voice input/output or text input/output.

4.4. Let's understand the Parameters-

For text input-

input_mechanism='text'

For voice input-

input_mechanism='voice'

For text output-

output_mechanism='text'

For voice output-

output_mechanism='voice'

For voice and text output-

output_mechanism='both'

5. What it can do (Features it supports)-

  1. Currently, it supports only english language
  2. Supports voice and text input/output.
  3. Supports AI based voice input (using whisper asr) and by using google api voice input.
  4. All intellectual task is process in JarvisAI server so there is no load on your system.
  5. Lightweight and able to understand natural language (commands)
  6. Ability to add your own custom functions.

5.1. Supported Commands-

These are below supported intent that AI can handle, you can ask in natural language.

Example- "What is the time now", "make me laugh", "click a photo", etc.

Note: Some features / command might not work. WIP. Tell me bugs.

  1. asking time
  2. asking date
  3. greet and hello hi kind of things goodbye
  4. tell me joke
  5. tell me about
  6. i am bored
  7. volume control
  8. tell me news
  9. click photo
  10. places near me
  11. play on youtube
  12. play games
  13. what can you do
  14. send email
  15. download youtube video
  16. asking weather
  17. take screenshot
  18. open website
  19. send whatsapp message
  20. covid cases
  21. check internet speed
  22. others / Unknown Intent (Premium Feature)

5.2. Supported Input/Output Methods (Which option do I need to choose?)-

You can set below parameter while creating object of JarvisAI-

jarvis = JarvisAI.JarvisAI(input_mechanism='text', output_mechanism='text',
                       google_speech_api_key=None, backend_tts_api='pyttsx3',
                       use_whisper_asr=False, display_logs=False,
                       api_key='99f605ce-5bf9-4e80-93a3-f367df65aa27')
  1. For text input-'

     input_mechanism='text'
    
  2. For voice input-

     input_mechanism='voice'
    
  3. For text output-

     output_mechanism='text'
    
  4. For voice output-

     output_mechanism='voice'
    
  5. For voice and text output-

     output_mechanism='both'
    

6. Future/Request Features-

WIP
You tell me
at info@jarvisai.in or my www.dipeshpal.in/social

7. Contribute-

  1. Clone this repo.
  2. Create your file in JarvisAI/JarvisAI/features/<your_file.py>
  3. Write entry function like this-
def some_func(*args, **kwargs):
    query = kwargs.get("query")
    entities = kwargs.get("entities")
    li = ['EVENT', 'FAC', 'GPE', 'LANGUAGE', 'LAW', 'LOC', 'MONEY', 'NORP', 'ORDINAL', 'ORG',
	  'PERCENT', 'PERSON', 'PRODUCT', 'TIME', 'WORK_OF_ART']
    topic = [entity[0] for entity in entities if entity[1] in li][0]
    return "This Code Done"
  • query is the text that is recognized by your microphone

  • entities Named Entity Recognition is a technique of natural language processing that is used for the categorization of the data

  • Example-

    query: who is Narendra Modi

    entities: [('Narendra Modi', 'PERSON')]

    topic: Narendra Modi

    So, now you got the topic from query and now you can play with the topic.

  1. Your function can return someting text or perform something. Return text will be automatically print / spoken by your system.
  2. In JarvisAI/JarvisAI/features_manager.py , import and add your function like this
try:
		from features.your_file import some_func
except Exception as e:
	from .features.your_file import some_func

action_map = {
	.....
	....
	'your_intent': some_func
}
  1. That's it now raise pull request. I'll verify your code. If working, ethical and all terms are followed, I'll approve.
  2. You will become contributer.

8. Contact me-

9. Donate-

Consider donating to JarvisAI to support our mission of keeping our servers running 24/7. Your contribution will enable us to continue doing great things and providing valuable services. Every little bit helps!

Click Here to support

Feel free to use my code, don't forget to mention credit. All the contributors will get credits in this repo.
Mention below line for credits-
Credits-

10. Thank me on-

License

MIT

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

JarvisAI-4.9.tar.gz (22.9 kB view hashes)

Uploaded source

Built Distribution

JarvisAI-4.9-py3-none-any.whl (30.8 kB view hashes)

Uploaded py3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page