Artificial intelligence (AI) technology that identifies defects by analyzing sounds, similar to how cameras assess product quality, is gaining attention. This technology distinguishes subtle differences in machine and fastening sounds that were previously discerned only by human ears, enabling the detection of anomalies in manufacturing environments.
DIPLI, an acoustic AI company, was founded in 2017 by CEO Lee Su-ji, a KAIST graduate with a master’s degree in brain and cognitive sciences from Seoul National University. After working as a researcher at Seoul National University Hospital and as a product manager in a healthcare company, she established DIPLI to develop acoustic AI technology.
The company’s core solution, 'Listen AI Industrial,' analyzes machine and environmental sounds in real-time on production lines to assess the quality of components such as motors, actuators, and bearings, as well as the fastening status of connectors and bolts.
Previously, skilled inspectors often relied on their hearing to determine the presence of defects, but Listen AI quantifies this process through data analysis. The company describes its solution as a way to standardize sound-based quality and fastening inspections that were previously dependent on individual judgment.
DIPLI has accumulated over 520,000 hours of acoustic data, exceeding 20 terabytes. This amount is equivalent to what 178 skilled inspectors would need to listen to for eight hours a day over a year. Based on this data, Listen AI analyzes the sounds of products produced on the production line in real-time.
Currently, the system evaluates more than 18,000 products per hour with 100% inline accuracy. The company reports that it analyzes over 150 million products annually, with the ability to assess the sound of a single product in just 0.1 seconds.
This technology also reduces the limitations of human hearing in inspections. It minimizes variations in judgment based on the inspector's skill level or fatigue and cuts the costs of installing and operating soundproof rooms by 60%. In manufacturing, poor fastening can lead to product recalls, creating a strong demand for quantitative inspection systems.
The solution is expanding its application, particularly in the automotive manufacturing sector. In May 2025, DIPLI began supplying its solution to a subsidiary of H Group in South Korea and Mexico, and in January of this year, it provided its technology to a global automotive motor manufacturer. Currently, it operates in 30 industrial sites worldwide, including H Group subsidiaries and global automotive motor manufacturers Hyosung Electric and Korail.
DIPLI is also pursuing expansion into overseas manufacturing sites. The company has signed an exclusive sales agreement for Listen AI with a local partner in the UK and is entering the European market. In North America, it is expanding its presence in the southern U.S. 'Battery Belt,' including Georgia, Alabama, and Tennessee, with plans to implement its solution in more than three manufacturing sites in the region by the end of the year.
DIPLI anticipates that as automation in manufacturing accelerates, the applications for acoustic AI will increase. As more factories employ robots for fastening connectors and bolts, the need for immediate assessment of both the fastening process and its results is growing. Consequently, the company is also working on expanding Listen AI into a multimodal AI that integrates vision inspection.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.