VoiceCoin was inspired by one of our group members, Michael, actually using and owning Bitcoin through the Coinbase Wallet. As various ways to look up the value of Bitcoin exist, as well as the "stock-like" infrastructure of the rise and fall of prices, we couldn't find a good way to personalize this information, and decided that voice integration through Amazon's Alexa would be much more user friendly.
What it does
VoiceCoin uses Alexa's voice recognition technology to interpret various requests by the user. It then, using APIs such as IBM's Watson and the Coinbase API, fetches meaningful information about your coins. From this real-time search, the app is able to relay us updates and information on Bitcoin, such as its current price, the percentage growth or decay per day, how much value your individual wallet balance has gained in value, and a vague price projection for the following day based on news headlines.
How we built it
Our application was built using Node.js, AWS Lambda, and the Alexa Skills Kit. Within this, we communicate with the Coinbase API for basic wallet functions, the Coindesk API for Bitcoin prices (current and historical), and various other sources to provide to the user. For our "projection", we start by getting news headlines about Bitcoin, from the Google News RSS feed. From this, we're able to combine Watson's APIs for sentiment analysis, emotional analysis, and language tone analysis to form a pseudo-accurate representation of the title's view of Bitcoin. From this, we weight these based on page rank, and combine them into a single value between -1 and 1 (also factoring in Bitcoin value growth from the past ten days). With these tools, we're able to provide a great experience to the user.
The largest challenge we came across was actually setting up Amazon Web Service so that we could write our own skills for Alexa. This was especially challenging because none of us had any previous experience with AWS or Alexa, so configuring a new skill while learning how to write a Lambda function was a challenging part of our project.
Another challenge we had to overcome was the use of the various Watson APIs, or getting enough entropy for a more accurate score. With only sentiment analysis, most articles returned 'neutral' while we identified them not to be, so we had to use other features of the API as well.
Accomplishments that we are proud of
Our largest accomplishment was actually getting a skill written in AWS for Alexa, so that certain voice cues and intents would communicate with our code correctly. Since this is the first hackathon for all three of our group members, we are proud to have a working product that works as designed.
What we learned
What's next for VoiceCoin
- OAuth2 authentication for Coinbase w/ VoiceCoin
- Google Assistant / Actions on Google support?