Natural Language User Interface Program with Python

Platform Supported

Reach Out

[![Ask Me | Anything ]

Prerequisites

• MacOS
  Tested on macOS High Sierra, Mojave, Catalina, Big Sur, Monterey and Ventura*

  • System Preferences → Security & Privacy → Privacy
  • Click + sign and add the preferred IDE and Terminal in the following sections in left pane.
    • Microphone - Required to listen and respond.
    • Accessibility - Required to use key combinations for brightness and volume controls.
    • Camera - [Optional] Required only during face recognition/detection.
    • Automation - Required to control System Events and other apps like Outlook and Calendar.
    • Files and Folders [OR] Full Disk Access - Required for all FileIO operations.

  ⚠️ Known Issue with pyttsx3 module on macOS Ventura 13.0: This version of macOS does not hold the attribute VoiceAge. Workaround has been raised as a PR
  
  

• Windows
  Tested on Windows 10

  • Settings → Privacy
    • Microphone - Required to listen and respond.
    • Camera - [Optional] Required only during face recognition/detection.
    • Unlike macOS, Windows pops a confirmation window to Allow or Deny access to files and folders.
  • Install Anaconda or Miniconda, and VisualStudio C++ BuildTools

Setup

Test Peripherals:

• Camera: camera.py
• Speaker: speak.py
• Microphone: mic.py
• Speech Recognition: recognizer.py

• Download the latest stable release from pypi or the latest un released version from github
• Navigate into the downloaded aegon or aegon-master directory.
• Run the following commands in a command-line/terminal:
  1. python3 -m venv venv - Creates a virtual env named venv
  2. source venv/bin/activate - Activates the virtual env venv
  3. which python - Validate which python is being used. Should be the one within the virtual env venv
  4. chmod +x lib/install.sh - Makes installation file as executable.
  5. bash lib/installs.sh - Installs the required modules based on the operating system.
  6. python jarvis.py - BOOM, you're all set, go ahead and interact with Aegon.

ENV Variables

Environment variables are loaded from a .env file and validated using pydantic

More on Environment variables

• ROOT_PASSWORD - System password to get the system vitals and run other sudo commands.

• TITLE - Title which Jarvis should address the user by. Defaults to sir

• NAME - Name which Jarvis should address the user by. Defaults to Aryan

• WAKE_WORDS - List of wake words to initiate Jarvis' listener. Defaults to ['jarvis'] (Defaults to ['alexa'] in legacy macOS)
  ⚠️ Jarvis has limitations on the wake words as it relies on ML libraries for wake word detection.

• VOICE_NAME - Name of the voice supported by the OperatingSystem. Defaults to the author's favorite.

• VOICE_RATE - Speed/rate at which the text should be spoken. Defaults to the value from pyttsx3 module. Typically 200

  To add more voices

  macOS:

  • System Preferences → Accessibility → Spoken Content → System voice → Manage Voices...

  Windows:

  • Settings → Time & Language → Speech → Manage voices → Add voices

• SENSITIVITY - Hot word detection sensitivity. Allowed range: [0-1] Defaults to 0.5

• TIMEOUT - Timeout in seconds until which the listener should wait for speech. Defaults to 3

• PHRASE_LIMIT - Timeout in seconds until which the listener will remain active. Defaults to None

• RECOGNIZER_SETTINGS - A JSON object that has with customized speech recognition settings.

  Custom settings for speech recognition

  These are customized according to the author's voice pitch. Please use mic.py to figure out the suitable values in a trial and error method.

  These settings are added (optionally), to avoid the hard coded PHRASE_LIMIT
  Cons in using hard coded PHRASE_LIMIT:

  • Disables the listener after the set limit even the speaker is actively talking.
  • Listener will be active until the set limit even after the speaker has stopped talking.

  Sample settings (formatted as JSON object)

  • RECOGNIZER_SETTINGS: '{"energy_threshold": 1100, "dynamic_energy_threshold": false, "pause_threshold": 2, "phrase_threshold": 0.1, "non_speaking_duration": 2}'

  Description

  • energy_threshold: Minimum audio energy to consider for recording. Greater the value, louder the speech should be.
  • dynamic_energy_threshold: Change considerable audio energy threshold dynamically.
  • pause_threshold: Seconds of non-speaking audio before a phrase is considered complete.
  • phrase_threshold: Minimum seconds of speaking audio before it can be considered a phrase - values below this are ignored. This helps to filter out clicks and pops.
  • non_speaking_duration: Seconds of non-speaking audio to keep on both sides of the recording.

• LIMITED - Boolean flag to run only the main version of Jarvis skipping background processes. Defaults to False Enforced based on the number of CPU cores.

• CAMERA_INDEX - Camera index that has to be used. Run camera.py to get the index value of each camera.

• DEBUG - Boolean flag to enable debug level for logging. Defaults to False

Features

• GIT_USER - GitHub Username
• GIT_PASS - GitHub Token
• WEATHER_API - API Key from openweathermap
• NEWS_API - API Key from newsapi
• MAPS_API - API Key for maps from Google
• BIRTHDAY - Birth date in the format DD-MM - Example: 24-April
• WOLFRAM_API_KEY - API Key from wolfram alpha.

Calendar/Meeting integrations

• ICS_URL - Shared calendar URL to get meetings information from. Should end with .ics
• EVENT_APP - To read events from outlook or calendar. Defaults to calendar
  💡   When calender is used, the name of the calendar within the Calendar.app should be Jarvis
  

Background scans [Defaults to 1 hour]

• SYNC_MEETINGS - Interval in seconds to generate meetings information using ics URL.
• SYNC_EVENTS - Interval in seconds to generate events information using calendar or outlook application.

Wi-Fi Controls

• WIFI_SSID - SSID of the wireless connection.
• WIFI_PASSWORD - Password for the wireless connection.
• CONNECTION_RETRY - Frequency in seconds to check for an active internet connection. Defaults to 10 seconds.

TV controls - Applies only for LGWebOS

• TV_CLIENT_KEY - TV's Client key. Auto-generated when used for the first time.
• TV_MAC - TV's mac address. Can be single [str] or multiple [list] mac addresses (to include both wired and wireless macs).

Car Controls - Applies only for JLR vehicles subscribed to InControl application.

• CAR_EMAIL - Email address to log in to InControl API.
• CAR_PASS - Password to authenticate InControl API.
• CAR_PIN - InControl PIN.

Garage Controls - Applies only for garages using MyQ garage controller.

• MYQ_USERNAME - Email address to log in to MyQ API.
• MYQ_PASSWORD - Password to authenticate MyQ API.

Telegram Bot integration

• BOT_TOKEN - Telegram BOT token.
• BOT_CHAT_IDS - UserID/ChatID for a particular user.
• BOT_USERS - Usernames that should have access to Jarvis.

OS Agnostic Voice Model

• SPEECH_SYNTHESIS_TIMEOUT - Timeout to connect to the docker container that processes text to speech requests.
  

  To enable independent speech-synthesis

  docker run \
      -it \
      -p 5002:5002 \
      -e "HOME=${HOME}" \
      -v "$HOME:${HOME}" \
      -v /usr/share/ca-certificates:/usr/share/ca-certificates \
      -v /etc/ssl/certs:/etc/ssl/certs \
      -w "${PWD}" \
      --user "$(id -u):$(id -g)" \
      rhasspy/larynx

  💡   Optionally run speech synthesis on a docker container for better voices but, response might be slower. If you don't have docker installed or simply don't want to use it, set the SPEECH_SYNTHESIS_TIMEOUT env var to 0. This is also done automatically if failed to launch a docker container upon startup.

---

Offline communicator

• OFFLINE_PORT - Port number to initiate offline communicator. Defaults to 4483
• OFFLINE_PASS - Secure phrase to authenticate offline requests. Defaults to OfflineComm
• WORKERS - Number of uvicorn workers (processes) to spin up. Defaults to 1

Stock Portfolio

• ROBINHOOD_USER - Robinhood account username.
• ROBINHOOD_PASS - Robinhood account password.
• ROBINHOOD_QR - Robinhood login QR code

API Features

• ROBINHOOD_ENDPOINT_AUTH - Authentication token to access the robinhood portfolio which is generated every hour.
• SURVEILLANCE_ENDPOINT_AUTH - Token to access webcam live feed via Jarvis API.
• SURVEILLANCE_SESSION_TIMEOUT - Session time out for /surveillance. Defaults to 300 seconds.
• STOCK_MONITOR_ENDPOINT_AUTH - Token to add a stock price monitor. (Will soon be made open-source)

Scheduler

• TASKS - Runs certain tasks at certain intervals.

  Sample value

  [
    {"seconds": 10_800, "task": "remind me to drink water"},  # Runs every 3 hours
    {"seconds": 21_600, "task": "turn off all lights"}  # Runs every 6 hours
  ]

• CRONTAB - Runs scheduled tasks using cron expressions without using actual crontab.

  Sample value

  [
    "0 0 * * 1-5/2 find /var/log -delete",
    "0 5 * * 1 tar -zcf /var/backups/home.tgz /home/"
  ]

Contacts

Aegon can send on demand notifications using a contacts.yaml file stored in fileio directory. Uses gmail-connector for SMS and email notifications.

Setup Instructions

Note: Jarvis currently supports sending emails only when the contacts.yaml file is present, however phone numbers can be used directly.

phone:
  Tony: 0123456789
  Thor: 1234567890
email:
  Eddard: ned@gmail.com
  Aegon: egg@yahoo.com

Smart Devices

A source file smart_devices.yaml is used to store smart devices' hostnames. Jarvis supports MagicHome lights and LGWebOS TVs.

Setup Instructions

Note: Jarvis currently supports only one hostname for TV but multiple for lights.

• The name used in the keys will be the identifier of those light bulbs.
• The source file (smart_devices.yaml) should be as following:

bedroom:
  - 'HOSTNAMES'
hallway:
  - 'HOSTNAMES'
hallway basement:
  - 'HOSTNAMES'
kitchen:
  - 'HOSTNAMES'
living room:
  - 'HOSTNAMES'
party mode:  # Light hostnames that needs to be engaged for party mode, if not present individual lights can be enabled
  - 'HOSTNAMES'
tv: 'LGWEBOSTV'

Automation Setup [Optional]

Jarvis can execute offline compatible tasks at pre-defined times without any user interaction. Uses an automation.yaml file as source which should be stored within the directory fileio

Setup Instructions

The YAML file should be a dictionary within a dictionary that looks like the below.

OPTIONAL: The key, day can be a list of days, or a str of a specific day or simply a str saying weekday or weekend when the particular automation should be executed.

Not having the key day will run the automation daily. Date format should match exactly as described below.

06:00 AM:
  day: weekday  # Runs only between Monday and Friday
  task: set my bedroom lights to 50%
06:30 AM:
  day:  # Runs only on Monday, Wednesday and Friday
  - Monday
  - wednesday
  - FRIDAY
  task: set my bedroom lights to 100%
08:00 AM:  # Runs only on Saturday and Sunday
  day: weekend
  task: set my bedroom lights to 100%
09:00 PM:  # Runs daily
  task: set my bedroom lights to 5%

Feature(s) Implementation

Please refer wiki for API usage, access controls, env variables, features' overview and demo videos.

Coding Standards

Docstring format: Google
Styling conventions: PEP 8
Clean code with pre-commit hooks: flake8 and isort

Linting

PreCommit will ensure linting, and the doc creation are run on every commit.

Requirement
pip install --no-cache --upgrade sphinx pre-commit recommonmark

Usage
pre-commit run --all-files

Pypi Package

https://pypi.org/project/jarvis-ironman/