My classmate burst into tears even before uttering a single word. The professor had called on her to deliver an impromptu speech for our required public speaking class at college. After the class I found out that it was her first time speaking in front of an audience! I was rather surprised that not all K-12 institutions gave such opportunities to their students, even though the Common Core Standards require our students by the end of middle school to be able to speak well in front of an audience (CCSS.ELA-LITERACY.SL.5.4). We decided to bring this problem to the Tech Crunch Hackathon with the hope of creating something innovative under the 24 hour time pressure.

Fortunately, there was an EdTech conference (EdSurge) taking place in Oakland this weekend, where we got to share our concerns with K-12 teachers and brainstorm a few ideas alongside them. One thing that stood out was the availability of standard assessments for every Common Core Standard except for speaking. Teachers can give math worksheets to assess algebra skills, or ask students to write an essay to assess literacy. But unlike a right or wrong answer in math, assessing speech is subjective due to its personalized nature, and thus no standard assessment is currently available to the teachers. Thus Articulate was born.

What it does

Articulate is an AI speech coach for K-12 students that can provide on-demand speech feedback. It also empowers students to speak more frequently by giving them a non-judgmental space to practice. Teachers don't have to install a new software as Articulate integrates into their existing Learning Management System (LMS). Here is the 6 step workflow for better speech assessments:

  1. Teacher assigns speaking prompt
  2. Student views assessment in the LMS
  3. Student clicks the record button to practice assigned prompt using the built-in webcam and microphone in a chromebook.
  4. Articulate listens ;-D.
  5. Student receives on-demand and personalized speech feedback.
  6. Student can practice again or submit the assessment.

This 6 step workflow not only saves teacher's valuable time, but also provides a scalable standardized method for assessing student speaking ability.

Feedback includes:

  1. Filler words
  2. Pausing
  3. Pacing

How we built it

Watson for audio processing. Golang for handling communication. Redis for data storage. HTML, CSS and Javascript for front-end.

Accomplishments that we're proud of

Sometimes finding the right problem to solve is the most challenging part. We were very fortunate to attend the EdSurge conference and find a real pain point facing the most instrumental individuals in our country: K-12 Teachers. I particularly remember one middle school teacher telling me that if we could build the things we brainstormed, we would be "changing lives of these students"!

What's next for Articulate

We hope to build new AI/ML models that have not yet been scaled commercially in order to provide more in depth speech feedback. Ultimately, our vision would be to become a data aggregator for K-12 speeches. Who knows what can be learnt from this big data set?!

Care about school? Want to create a far better K-12 experience for future students? Please come and help us brainstorm on how we can take Articulate to the next level.

Built With

Share this project: