FPT secures US patent for AI-powered sound detection technology

On Monday, the firm revealed that its neural network-based system, SoundAI, has received U.S. patent approval.

The technology is designed to help businesses, especially in manufacturing, address key challenges in industrial maintenance and sound monitoring.

By recognizing unusual sound patterns, the system can detect equipment malfunctions, operational errors, or security risks.

Thanks to its extensively trained encoder-decoder neural network, SoundAI can accurately identify anomalies even in noisy environments.

With abnormal sound detection accuracy reaching up to 95 percent, the system enables companies to proactively maintain equipment, monitor product quality, reduce reliance on manual inspections, and improve operational efficiency.

FPT has integrated this advanced AI model into its Intelligent Inspection (I2) SoundAI solution, which is currently being deployed for clients in the manufacturing and automotive sectors.

“This patent further demonstrates FPT’s capability to develop proprietary technologies that solve complex real-world problems and meet stringent international standards in AI research and application,” said Nguyen Xuan Phong, director of AI at FPT Software.

Earlier this year, FPT became the first Vietnamese company to secure a U.S. patent for a generative and code-ranking neural network system.

That system is now being applied in AI4Code, FPT’s intelligent programming assistant, to enhance employee productivity and improve product quality for clients.