Commend: introducing 
deep neural networks

During 2024, Commend introduced Deep Neural Networks, to significantly perfect speech intelligibly and optimize the signal processing of its intercom terminal series. Developed together with the University of Art and Design Graz, these deep neural networks are highly efficient network algorithms for adaptive Artificial Intelligence, ensuring that the voice signal in intercom systems is perfectly cleaned of ambient noise and echoes during transmission. Any residual noise is almost completely suppressed. The neural network can react with high precision to the many different challenges of everyday life - for example different voices, ambient noises and situations.

The result is an unprecedentedly clear, intelligible sound profile that corresponds to transmission in a noiseless environment. In emergency situations such as an accident on the highway or in a production hall, every second counts. If background noise or echo effects distort the spoken word or make it completely unintelligible, in the worst case lives are at stake.

Problem-specific neural network structures are trained using standardized machine learning algorithms. Using countless hours of audio data, they are systematically 'taught' the desired recognition capability.

At the 2024  international security trade fair "ICS West" in Las Vegas, the US Security Industry Association honored the neural network product with the prestigious "Best in Communication and Networking Solutions" award.