74% of American's suffer from speech anxiety. The top 3 phobias in the US are #3: Spiders, #2: Death, and #1: Public Speaking. Practice makes perfect and the age old practice technique is to practice in front of a mirror. What if that mirror could tell you when you said filler words?
What it does
Our app functions as a mirror 2.0. Any time you say a filler word (like, um, but, so, alright) our app makes a gentle beeping sound to notify you of your speaking taboo. After completing a practice speech, you are brought to a page that visualizes data extracted during your speech. We track words per minute over time, average words per minutes, speech duration, number of filler words, and number of total words.
How we built it
The main app is for iOS using a mixture of Swift and Objective C. The natural language processing was done exclusively with the Bing Speech API as part of Microsoft Cognitive Services. Data visualization is done on the web with Node and Angular and is hosted with Azure. We connect the two using MongoDB. The web based visualization is also integrated seamlessly into the app.
Challenges we ran into
We had a lot of trouble with APIs. Some APIs would only work in Objective C while others needed Swift. We were able to incorporate both in our projects but there were many merge conflicts we had to struggle through.
Accomplishments that we're proud of
It actually works! You can record yourself speaking and the app beeps to detect filler words in real time with surprising accuracy.
What we learned
We definitely learned more about Swift and Objective C as well as how to better work with APIs.
What's next for Articulate
Our next step is to integrate a community aspect into the app. We envision a way for anyone to upload their speech and get comments and helpful critiques from those they are close to. We envision building a community of proud public speakers.