How did we set SAIL at the Banyan Hack ‘16?
The Banyan Hack 2016 was a 24-hour Hackathon that invited students from all over the country, but mostly Bengaluru, to work on a hack that dealt with Machine Learning and Artificial Intelligence systems, and we were a part of the action too!
This hackathon was actually supposed to be held on the 10th of September, but was postponed due to the Karnataka Bandh, and we were truly disappointed that we wouldn’t be hacking that day, but we got a lot more time to think about a solid idea and really start to love it.
Teaser
Apple came up with it’s 107-second Don’t Blink ad, and as an inspiration from it’s artistic beauty, we came up with our own 40-second teaser to our hack.
The Idea
Our idea was conceived in the last few days of the week, just before the Hackathon, and it was something that all of us loved to work on the second we heard about it. SAIL was brought to life as just an idea.
SAIL (Speech Assisted Interface Library) is an Android library that can integrate with your developing or on-the-shelf application and provide a speech interface layer over your existing interface. It will allow a lot more accessibility to the end user, by literally letting him/her converse with the application.
The Implementation
As a developer, all you need to do is extremely simple. Just add the library dependency, create an object of our Sailboat class and call initialize() which takes two parameters: the current Activity’s context, and an ArrayList of views in the Activity. That’s about it. You’re set to go.
We use the magic of Machine Learning and Neural Networks to process the natural language and make the system train better with natural speech recognition. We have also structured our data so beautifully that the more others use the library within their application, the more it will benefit all users and improve accuracy, all the while maintaining complete abstraction and security for the developer’s and user’s data.
To showcase this, we built a simple Cab Aggregator application to mimic those on the market today, so that we could easily summon a taxi with our voice, from any screen.
While on this screen, all we had to do was talk to this application. We set our hot word as Candy, so all we had to say was:
“Candy, book me a cab that will pick me up from Marenhalli and drop me off at Indiranagar…”
and it would present me with a booking confirmation, where I can further converse with my application by saying
“Candy, call the driver”
The Team
The Hackathon team was restricted to 4 people, but we have a lot of people to thank in the process. Let’s begin with our own team.
Darshan N. & Rohan K. — The Android Duo. They were responsible for all the development regarding the library, and database structuring.
Chandramouly K. — The Machine Learning Baba. He was the one-man-army who handled everything on the Machine Learning and the Django REST API portion of it all. Needless to say, he went away with a lot less hair than he came in with.
Raja Rajan A. — The Designer, yours truly. Now you would wonder… what would a designer possibly do in a Machine Learning hackathon? Well, I had the same doubts, but we used that oddity to our advantage and implemented a lot of clean design concepts in our dummy apps, and created the video teaser that you all saw above, and if you didn’t: scroll UP!
We would also want to thank our HoD & Vice Principal, D. R. Ramesh Babu Sir, our friends, Abhishek Kori and Rohan Antony and everyone who wished us good luck and kept us awake with their incessant messages.
The Food
This is always a driving factor for me when it comes to Hackathons. If there’s good food, an uncrowded coffee machine, I’ll shift my office to the pantry. Unfortunately, the food at this hackathon was not that properly managed. We were given Rice and Rajma Chawal, Chappathi, Puri and Sweets. The menu was decent enough, although it’s not really ‘hackathon food’, but the main issue was that none of the food was served in time.
So we took issues to our own hands, and went out on frequent breaks to go shopping at the local provision store. In fear of getting hungry at midnight, we robbed the store of all their delicious treats and bought everything we could set our eyes on. All that food is now history, for now.
The Experience
It has been quite a long time since all of us attended a hackathon, and these sleepless nights were something that we wanted desperately. To be surrounded by intelligent minds, discussing math and machine, it was the environment to be in, and at the NASSCOM 10K Startup Office no less.
We got to meet other people who were working on really innovative projects, and even old friends that we had made in previous Hackathons. All the while, assessing in our heads, hoping that our idea beat their idea. In the end, it’s all about the development, not who wins. (being diplomatic, of course)
When we came in with the idea of SAIL, we had no clue what we were going to build… but only as time went ahead and we started to etch out our plan on the tables and walls, we grew an understanding and finally, SAIL sailed away.
The Results
We’re just as nervous as you are to know… the results aren’t out yet, but we are sure whatever team wins, they deserved it a lot :)
Stay tuned and we’ll update the post here as soon as we know what the results came in as.
Update
We regret to inform you that…
