Dragonfire

the open-source virtual assistant for Ubuntu based Linux distributions

Special thanks to Jassu Ilama for the beautiful 3D modelling and material design of this avatar.

Dragonfire goes through these steps for each one of your commands, respectively:

• Search across the built-in commands and evaluate the algebraic expressions
• Try to Learn using Advanced NLP and Database Management Techniques
• Ask to Open-Domain Question Answering Engine (Searches Wikipedia for an answer)
• Respond using the Deep Conversation system, a seq2seq neural network trained with Cornell Movie-Dialogs Corpus

Dragonfire uses Mozilla DeepSpeech to understand your voice commands and Festival Speech Synthesis System to handle text-to-speech tasks.

Feel free to join our Gitter chat room. You can also directly talk with Dragonfire herself via her Twitter account.

Supported Environments

Installation

To run Dragonfire on a desktop Debian or Ubuntu system, either download the latest release (the .deb file) and install as follows:

sudo dpkg -i dragonfire_1.1.1_amd64.deb

or clone the GitHub repository and run

sudo make install

in the repository directory.

To install the dependencies, run sudo apt-get -f install right after the dpkg -i command. The installation will automatically download the pre-trained English model of Mozilla DeepSpeech (1.31 GB download size) and will place it under /usr/share/dragonfire/deepspeech/models directory. You can manually download the model if you wish.

If you want to run Dragonfire on a server, you should install the Docker image (which does not install the huge DeepSpeech model required for speech recognition):

docker pull dragoncomputer/dragonfire

Usage

usage: dragonfire [-h] [-c] [-s] [-j] [-v] [-g] [--server API_KEY] [-p PORT]
                  [--version]

optional arguments:
  -h, --help            show this help message and exit
  -c, --cli             Command-line interface mode. Give commands to
                        Dragonfire via command-line inputs (keyboard) instead
                        of audio inputs (microphone).
  -s, --silent          Silent mode. Disable Text-to-Speech output. Dragonfire
                        won't generate any audio output.
  -j, --headless        Headless mode. Do not display an avatar animation on
                        the screen. Disable the female head model.
  -v, --verbose         Increase verbosity of log output.
  -g, --gspeech         Instead of using the default speech recognition
                        method(Mozilla DeepSpeech), use Google Speech
                        Recognition service. (more accurate results)
  --server API_KEY      Server mode. Disable any audio functionality, serve a
                        RESTful spaCy API and become a Twitter integrated
                        chatbot.
  -p PORT, --port PORT  Port number for server mode.
  --version             Display the version number of Dragonfire.

or with Docker: docker run dragonfire [-h] [-c] [-s] [-j] [-v] [-g] [--server API_KEY] [-p PORT]

or simply start from your Linux application launcher.

To activate Dragonfire say DRAGONFIRE or HEY or WAKE UP.

To deactivate her say GO TO SLEEP.

To silence her say ENOUGH or SHUT UP.

To kill her say GOODBYE or BYE BYE or SEE YOU LATER or CATCH YOU LATER.

:warning: Facing with a problem? Take a look at the Troubleshooting:hammer_and_wrench: section.

Built-in Commands

Dragonfire DEVLOG #3 - Built-in Commands

DRAGONFIRE | WAKE UP | HEY
GO TO SLEEP
ENOUGH | SHUT UP
WHO AM I | SAY MY NAME
MY TITLE IS LADY | I'M A LADY | I'M A WOMAN | I'M A GIRL
MY TITLE IS SIR | I'M A MAN | I'M A BOY | CALL ME *
WHAT IS YOUR NAME
WHAT IS YOUR GENDER
FILE MANAGER | OPEN FILES
OPEN (BROWSER | CHROME | FIREFOX)
PHOTOSHOP | PHOTO EDITOR
INKSCAPE | VECTOR GRAPHICS
VIDEO EDITOR
OPEN [CAMERA, CALENDAR, CALCULATOR, STEAM, BLENDER, WRITER, MATH, IMPRESS, DRAW, TERMINAL]
SOFTWARE CENTER
OFFICE SUITE
KEYBOARD *
ENTER | NEW TAB | SWITCH TAB | CLOSE | GO BACK | GO FORWARD
SCROLL LEFT | SCROLL RIGHT | SCROLL UP | SCROLL DOWN
PLAY | PAUSE | SPACEBAR
SHUT DOWN THE COMPUTER
GOODBYE | BYE BYE | SEE YOU LATER
(SEARCH|FIND) * (IN|ON|AT|USING) WIKIPEDIA
(SEARCH|FIND) * (IN|ON|AT|USING) YOUTUBE
(SEARCH|FIND) * (IN|ON|AT|USING) (GOOGLE|WEB)
(SEARCH IMAGES OF|FIND IMAGES OF|SEARCH|FIND) * (IN|ON|AT|USING) (GOOGLE|WEB|GOOGLE IMAGES|WEB IMAGES)
WHAT'S THE TEMPERATURE IN *
WHAT TIME IS IT

Learning examples

Dragonfire DEVLOG #4 - Learning

Open-Domain Question Answering (ODQA) examples

DeepPavlov SQuAD BERT model based, a reader for the Wikipedia articles that are retrieved from web, found by the subject

Deep Conversation examples

Supported Distributions: KDE neon, elementary OS and Ubuntu. All modern releases (Ubuntu 12.04 LTS and above) of these distributions are fully supported. Any other Ubuntu based distributions are partially supported.

Contribute

If you want to contribute to Dragonfire then please read this guide.