A new way to monitor infant language development using Artificial Intelligence

Industry
Artificial Intelligence
Timeline
May 2019
Output
Mobile app
Office
Rangle Toronto

Highlights

  • From idea to proof of concept in just a few weeks
  • Enables parents to improve learning outcomes for their young children
  • Babbly team closed a successful pre-seed investment round with machine learning PoC

The challenge

Babbly estimates that one in ten children have difficulty with their language development. Many of these children remain undiagnosed and untreated as they enter pre-school, leading to long-term difficulties with learning. Babbly’s founder Maryam Nabavi wanted to improve language development outcomes in a way that would be simple and accessible for busy parents.

The core problem

Helping parents understand and measure their child’s development could mean children receive the support they need before they enter the school system. To build their app, the Babbly team first needed a way to identify vocalization from both babies and adults.
Above: Maryam Nabavi (CEO) and Shane Saunderson (Co-Founder) from Babbly and Ryan Marchildon and Jan Scholz from Rangle.

The process

In two weeks, our AI team built a proof of concept machine learning algorithm for Babbly. The algorithm identifies infant vocalization events such as crying and babbling, and can distinguish them from adult speech. This speech recognition is the foundation for Babbly’s AI-powered app for parents, allowing them to collect and interpret data on their child’s cognitive development. The app tracks a child’s development within the spectrum typical for their age group.
Above: The proof-of-concept demo in action, showing speech classification on audio clips never previously seen by the algorithm. The demo was used by Babbly to support investor pitches and demonstrate the viability of Babbly's concept.

Lean AI:

Using a lean approach, Rangle was able to deliver a working algorithm, together with a browser-based UI for a demo, in just a few weeks. The team synthesized a training dataset to reduce data labelling time, a potentially time-consuming process. As part of our values and our commitment to Babbly, the Rangle team also provided a seamless handoff with code assets for easy updating and retraining of the AI model. This has allowed the Babbly team to make improvements and iterate on their initial concept. Learning from the groundwork of our expert AI team, Babbly’s developers have a framework for processes and delivery that keeps them focused on outcomes.
Working with Rangle helped us avoid a lot of the early mistakes we could have made and put us on a fast track towards launching our product.
Maryam Nabavi
CEO and Founder of Babbly

The outcome

Babbly has built their AI-powered speech recognition and learning app. Rangle’s support has helped Babbly accelerate the growth of their internal machine learning practice, and do additional model development independently. With our partnership, Babbly’s team has developed the necessary skills to grow and expand their work and their app.
Above: Maryam Nabavi accepts the people’s choice award for best pitch at the ElevateR Pitch competition at Elevate Toronto.
With the help of the demo from Rangle, Babbly:

More on digital transformation

Learn about what we offer