Human Talk

We give non-verbal students unique, human-like voices so they can speak with clarity, identity, and dignity — not robotic sameness. Our AAC app makes communication personal and empowering.

Comment

Inspiration

Many non‑verbal students on the autism spectrum rely on communication apps like TouchChat to express themselves. Unfortunately, these tools use outdated, robotic‑sounding voices that often mispronounce words and make communication harder for both the user and the listener. Even more importantly, every user is forced to share the same generic voice, leaving them without a sense of individuality or identity when they speak.

Research in Speech Pathology shows that this lack of natural speech can severely hinder language development. In some cases, students graduate without ever learning to communicate effectively — not because they lack the ability, but because the tools failed to support their growth.

What it does

Our goal is to change that. We want to give every person who depends on a communication device the ability to sound unique — with a voice that reflects who they are, not a one‑size‑fits‑all synthetic voice. By integrating Higgs Audio and Eigen AI to incorporate human-like voices, our application aims to modernize AAC technology, making communication more natural, easier to comprehend, and more motivating. This helps users practice language more often accelerating communication skills and enriching social interaction.

How we built it

Python, Eigen AI, Higgs Audio TTS, REST, Flask, HTML, CSS, JS

Challenges we ran into

Figuring out how to generate unique voices that don't change due to entropy. This solution is found in our presentation slides.

What's next for Human Talk

Take the app to the next stage from prototype to production, add images to the words. Change this from a web application to a mobile application for ease of use.

Once the first step is done, then we will test this app in schools and see how students interact with it. Based on their feedback make changes necessary.

On the onset, we will communicate with Eigen AI to work on reducing prices for API calls. Current charge is 10 cents per audio minute. This needs to be drastically reduced to make this app even feasible for an individual with verbal difficulties to use as their main method of communication.

Built With

Share this project:

Updates