LittleBabble

Intro screen
Home screen
Listening screen

Inspiration

One of the most key sets of developmental milestones is a child’s language development. Delays or other issues could be important signs of developmental disorders and possibly lead to academic and social difficulties, learning disabilities, and behavioral problems. Therefore, it is critical for parents to be aware of their child’s progress and the expectations that should be met during the stages of growth. While parenting is never easy, picking up on or understanding the stages is significantly easier for economically stable or able-bodied parents, who have plenty of resources at their hands.

What it does

Some apps available today include ChatterBaby, which uses AI to analyze a baby’s cries. We want to take it one step further, using a language recognition algorithm in order to analyze audio clips from babies - young children and determine which stage of language development they are in. Some stages include babbling (ma-ma, ba-ba) and holophrases (one word sentences that express complex ideas). This way, parents can use our app to gauge how their child is growing, and see if they should be consulting a doctor for further investigation. This will be critical for catching cases like autism, hearing impairments, learning impairments at a much earlier age, and assist parents to be better prepared to support their children. Our app will be hugely beneficial for deaf parents, who have a harder time picking up these vocal cues. It will also benefit minority parents, as children of color are typically diagnosed with autism or other disorders at a much later age than their white peers. We want to close these disparities so all parents can be properly informed and make the parenting process easier for everyone.

How we built it

CMU Sphinx is a set of speech recognition development libraries and tools that can be used in speech related applications. We utilized this to create a program that could recognize words from an audio input. For the app itself, it was developed in Swift.

Challenges we ran into

The learning curve for CMU Sphinx was extremely steep, and we found ourselves not able to develop a specific model in time. In the future, we would build and train a language model for the different stages, and be able to analyze from a live audio feed rather than just audio files. It was also difficult to merge the speech recognition feature with the app itself. We both have very little experience in app development and language analysis.