slp.ai

Home Page
Test Page
Result Page

Inspiration

Speech language pathologists (SLPs) are medical professionals that offer speech therapy to individuals who experience speech and language disorders. A typical initial SLP assessment is broken down into several sessions due to the volume of tests needed to fully assess the person's possible speech and language disorder. What if there is a program to screen individuals before taking an initial SLP assessment?

What it does

slp.ai helps streamline the assessment process by giving an initial screening test. This program focuses on testing elementary-aged (3-7 years old) children on their cognitive level and their ability to speak in a coherent manner. The test asks for a child to describe an image and returns a probability of how accurate the child's description is. Based on the result, a SLP can analyze whether or not further testing is necessary.

How we built it

In the test, the child is asked to describe an image. The program analyzes the recording and turns it into text. Then, we use Co:Here's classify algorithm to check the accuracy and coherency of the recording sample. This algorithm returns the relevance of the child's response, which an SLP can use to determine if an assessment is required.

Challenges

Some challenges that we faced include learning how to use python’s speech recognition module and Google Cloud’s speech-to-text services. We manually built test cases for Co:Here’s clarify algorithm so that it can intelligently learn how to determine the accuracy of a child’s description of the image. Our main difficulty was the high learning curve when building and designing front-end and back-end, figuring out how to merge both of them, and incorporating all the different programs into the web application.

Accomplishments

We learned so much about how to create a fully functional web application. Coming from a team with little to no experience in web development and what even a hackathon was, we are so proud of what we built in the little time that we had.

What we learned

We learned

how to use front-end languages and tools, such as React.js, HTML, CSS, etc
how to delegate segments of the project
how to think outside of the box to find creative solutions to tough problems
how to work together as a team

What's next for slp.ai?

Our hope is to further develop slp.ai so that there will be more types of tests available, including sections of an actual SLP assessment, such as the OWLS test. Since our program only targets the 3-7 years old age range, we also hope to expand to include initial screenings for other age groups.