OtoSense Development of a Sound Recognition App

The solution

The first product developed by the team is called OtoSense, based on a technology called SIR, for Sound Intelligent Recognition.

OtoSense is a sound recognition app running on various mobile platforms. It records and analyzes background sounds and sends notifications to its user if any noticeable sound is recognized (fire alarm, doorbell, etc.). It also allows users to record specific sounds they want to be noticed about. The product’s specific user interface is adapted to people who might as well suffer from visual disorders.

How does this technology work?

Alike many AI-based systems, the OtoSense sound recognition app has two main components:

• The AI engine which implements different techniques from signal processing to deep learning algorithms and that may be trained to learn about sounds to recognize (alarm, baby crying, engine wearing off, house sounds, etc);
• The edge devices that process chunks of noise, extract the sound features of interest, and trigger alarms in the form of visuals, emails, SMS, notifications, etc. The edge devices may be either smartphones/tablets (iOS, Android), wearables such as smartwatches, electronic boards (e.g. Raspberry PI), or web applications.

OtoSense is an innovative solution, far more efficient and less expensive than the preceding systems.

Pentalog set up a dedicated team of 5 engineers based in Chisinau (Moldova) to support the development of the solution. They first worked on the improvement of the first prototype and then took over the development of the iOS and Android versions of the application, and its connection with the Pebble smartwatch.