Engineers from Ambient System in collaboration with researchers at Gdansk University of Technology in Poland, have developed a system that automatically adjusts the tone, level and pace of publicaddress
messages. Powered by two proprietary algorithms, it is already deployed across railway, metro and airports networks in multiple countries.
It is already deployed in demanding environments such as the Delhi Metro - one of the world’s busiest and noisiest transport hubs, serving millions of passengers every day- as well as across railway and metro networks in Poland, Turkey, Sweden and Finland.
“Our aim was to create a system that not only meets rigorous international standards but, above all, improves comfort and the sense of safety for people in public, commercial and industrial facilities. The smartVES PA/VA was developed through close collaboration between Ambient System engineers and the scientific team at Gdansk University of Technology. It demonstrates how effectively commercial and academic expertise can be combined,” said Marcin Starzynski, CEO of Ambient System.
“We implemented two proprietary digital audio-processing algorithms. Adaptive filtering continuously adjusts announcements to surrounding acoustic environment, ensuring speech remains intelligible despite background noise. The speech time-transposition algorithm analyses the operator’s voice and corrects speech tempo in real time—even during live announcements when a speaker may stumble or speak too quickly under stress.
As a result, messages remain clear even in large spaces with significant reverberation,” added Prof. Józef Kotus of Gdansk University of Technology.
Two algorithms, that are critical to ensuring announcements remain clear, underpin smartVES performance: the speech time-transposition algorithm and the adaptive filtering algorithm.
The first dynamically adjusts playback speed to the acoustics of the space, including live operator messages so rushed or hesitant speech stays intelligible. The second assesses the environment’s acoustic profile and existing background sounds, then reshapes the speech signal so announcements cut through background noise by emphasizing frequency ranges less affected by interference.
image: shutterstock/aquatarkus