Currently I'm not working on this project. But I'm planing to get back to it, refactor old/ugly/hack-it-just-works... code and maybe change the direction of project a little bit. IDK when though, maybe this month. (wrttten in June 2019)
Simple robot with eventual aspirations for robot. Project is build with specifically Raspberry Pi in mind (see Syl-RaspberryPI repo). It listens and speaks.
Currently project uses:
- HTML - structure of the web page, currently the only way to use the app
- JavaScript - no front-end frameworks or libraries. App uses speech recognition and speech synthesis (more on that later). Node.js and Grunt.js for easier deployment and to run it on final platform (Raspberry Pi). This list will be longer as the project will get more complex.
To check if your browser supports speech-to-text run this in your browser console:
var msg = new SpeechSynthesisUtterance('Hello World');
window.speechSynthesis.speak(msg)
If you get speech output then it's supported. Web Speech API has support for all the latest Chrome/ium browsers. More on that here. I'm aware it's not currently working on mobile ;(
Live preview: https://kubami9.github.io/Syl-Web/
This project is currently in its alfa, which means there are many things to do and some of them will set the path for further ideas. If you would like to become a part of it you can check out issues and if there's something that's not assigned to me you can feel free to write a comment and take it. There won't be so many free issues at the very beginning as I want to make some things on my own so the project can take a path that I have in mind.
To run the project you need to have Node.js installed. If/after you do it, you only need to install Grunt.
To install Grunt CLI run npm install -g grunt-cli in your node terminal. Then you can follow this 3 points and you should be good.
Use grunt command to update built.js file (which is main.js and features.js combined into one). It will fire contrib-concat plugin that concats both these files.
dist/
built.js
Gruntfile.js
LICENSE
README.md
features.js
index.html
main.js
package-lock.json
package.json
index.html - web page structure
main.js - speech recognition and synthesis and features array with keywords Syl recognizes
features.js - features, answers Syl can give when she hears a specified keyword
Gruntfile.js - just a Gruntfile, for more info check out this
dist/build.js - main.js and features.js files combined with Grunt
Add features to the features.js file and preced them with meaningfull comment. Variable with answer Syl is going to give should end with Info. So for example for weather feature it should be weatherInfo and for music feature musicInfo and so on. Once your feature is written add keyword that you want Syl to understand in the the keywords array in main.js file and add your variable to the answers array (in the same file).
NOTE: keyword position in keywords array needs to match answer position in the answers array.
By human phrases I mean things that humans say but aren't neccessarily informative, like "hello" or "how are you?". Add your phrase(s) to humanPhrases object in features.js file. Note that phrases are stored in arrays starting with phrases prefix, like phrasesGreetings and answers are stored in arrays with answers prefix, like asnwersGreetings. Those pairs need to match for easier development. If you added new arrays to humanPhrases object go to main.js file to speech synthesis section and checkForCommands() function and add your own loop that checks for your phrases, you can use an existing loop as an example.
Currently Syl can tell you:
- what's the time
- what's the date
- her name
- who's her creator
- she has no clue what are you talking about and maybe something else. I don't know it's still under intense development! To see more stuff I'm planning visit issue about new features.
TL;DR: Adding more options, fixing bugs, improving/adding algorithms. Migrating for Raspberry Pi (see Syl-RaspberryPi repo).
My main focus right now is to make this project specifically for Raspberry Pi. Although I also want it to work in the browser kinda like it works now but with more options and chat. I'm planning to add much more features and commands and to keep doing it. I think that's how most of the contributions will look like after the core is ready. I don't wanna use any framewokrs or libraries (or at least as few as possible). I also would like to make a version for mobile phones (android) but this might be a veeeery far future. I'm open for some cool ideas.
E-mail me at: [email protected]