eKnock

Inspiration

Mental health is an issue coming to the forefront of society. According to CAMH, Canada loses $51 billion due to health care costs, reductions in health-related quality of life, and lost in productivity. Our app, eKnock, was created to reduce this burden by using music to help identify and express ones emotions. It also provides a link to the Canadian Mental Health Association's website so that the user can seek help if need be.

What it does

eKnock prompts users to have a quick conversation. The app then analyzes the content of the converstaion and determines the users mood. Based off of this mood, it recommends a song on Spotify and sends a link to it.

How we built it

We built this app in Android Studio. The speech-to-text feature was a preexisting feature in Android Studio. We used IBM Watson to analyze the tone of the text. To calculate a valence value, which is how positive or negative a song is (a rating between 0.0 - 1.0), we devised an algorithm using the mood and the intensity value of the mood outputted by the API. To use the Spotify Recommendations api, we first needed to authenticate the app and then the user. The app was authenticated on Spotify's developer dashboard, and the user signs into Spotify to start the app. The valence value is used to get the link to a song from the Spotify Recommendations api. The user can go to the song by clicking a button on the screen.

Challenges we ran into

Like with most hackathons, implementing our APIs was the greatest challenge: specifically, the Spotify authentication apis. In order to get an authentication token, we had to generate a code that would generate and authentication token and a refresh token.

What's next for eKnock

From a technical standpoint, we would like to have an embedded Spotify audio player. In addition, eKnock could prompt the user with questions based off of their mood and verbally respond back. We also wish to utilize voice and speech pattern sentiment analysis to analyze the mood more accurately. Finally, we would like to refine the Spotify recommendation search by including more parameters, upon further research. Additionally, we would like to do more research into AI so that eKnock can also help alleviate mental illness by being a friend that supports the user and caters to their needs based off of the mood analysis.

Built With

android-studio
ibm-watson
java
spotify-recommendations
spotify.sdk.android.authentication

Submitted to

StarterHacks 2019

Created by

I implemented the Spotify APIs and returned a link to the recommended song based on the received valence value. I also helped determine how to calculate a valence value based off of the IBM Watson's output.

Kaitlyn Lee
The majority of my work revolved around implementing the IBM Watson Tone Analyzer API into our application. The function needed to take in a sentence and return a value based on an algorithm I created that would be passed on to the Spotify API to generate a link to a song. Furthermore, I worked on app design and structure, as well as the implementation of the Spotify API. Finally, I assisted my team members with the development of the UI.

Jerry Tan
I worked on researching Spotify API's that helped process songs based on the the user's mood. I learned a lot about different API's and how to implement them into programs like Android Studios.

Calvin Dong
I worked on the front end, mainly working on design and the user interface. I created all the app's images and icons by drawing them by hand on Microsoft's Paint 3D. I had also learned some Android Studio so I could help implement the pictures into the app.

Enoch Fok
I implemented the speech to text feature of the application, allowing the data received to be processed by IBM Watson and finally return a recommended song from Spotify based on the user's mood. Also, I was the main developer of the app design, taking responsibility of creating the UI and making it look aesthetically pleasing. Finally, I also aided my other team members by helping them implement the IBM Watson API and the Spotify API.

Ivan Chow