How Nestlé Uses Sound-Based Machine Analytics to Prevent Downtime

The project, named "Listen to Machine", was an experimental project developed for Nestlé to improve factory machine monitoring. Using Machine Learning, Visium built a system to listen for anomalies in the machines' operation and alert the operator when necessary, improving efficiency and reducing maintenance costs.

The Client
Visium and Nestle Case Study
The Project

Main challenge

To prevent machine interruptions and breakdowns on their manufacturing sites, Nestlé’s operators had to monitor the production line continuously -  a time consuming and costly operation.

The company was ready to explore a more efficient and accurate way to detect machine anomalies and maximize uptime.

How we helped

Most operators rely on alarms triggered by a machine’s sensors to identify malfunctions. But these are legacy machines, not equipped with any sensor technology. So, how can we create a sound-based system that would notify operators in real-time about potential issues while saving them a trip? 

We developed a device equipped with a directional microphone that uses an ML model performing edge computing. The solution collects data from the microphone to track sound patterns in real-time, detects anomalies, classifies and escalates them to the operator’s dashboard for timely intervention and further review - all in less than 1 sec.

The Impact


Fast and accurate interventions would help prevent machine failures, have fewer disruption, and improve throughput.


The life expectancy of the machines would increase with less need for maintenance work.


As operators are notified on their dashboard in real-time they could quickly assess the anomaly from the control room.

Check this video for more information on the project.