“

Miroslav Trinko Geek, specialty programmer, but journalist by profession. Rider, tennis player and fan of Formula-1. I write about technologies, smartphones and electric vehicles.
The research is based on the adaptation of an open model of Whisper recognition, which has been transferred to radar data, which helped to improve the quality of speech recognition, even though these data are very noisy and poor. The radar was installed at a distance of about three meters from the phone, and the data obtained processed artificial intelligence, which could be recognized up to 10,000 words.
Although the accuracy of the system is not ideal, even partial recognition of keywords can have serious consequences for privacy. Researchers compare the work of their technology with reading on the lips – this technique also does not recognize all the words, but helps to understand the essence of the conversation, especially when considering the context.
It is important that this technology is concerned about possible threats to privacy, as attackers can use it for remote listening to personal conversations without the user’s knowledge. The researchers themselves emphasize that their work aims to warn society about such risks and promote the development of means of protection against this type of espionage.
The study is supported by the US National Scientific Fund, and the team plans to continue working on the protection of personal information from the latest technological threats.
”, – WRITE: mezha.media

Miroslav Trinko Geek, specialty programmer, but journalist by profession. Rider, tennis player and fan of Formula-1. I write about technologies, smartphones and electric vehicles.
The research is based on the adaptation of an open model of Whisper recognition, which has been transferred to radar data, which helped to improve the quality of speech recognition, even though these data are very noisy and poor. The radar was installed at a distance of about three meters from the phone, and the data obtained processed artificial intelligence, which could be recognized up to 10,000 words.
Although the accuracy of the system is not ideal, even partial recognition of keywords can have serious consequences for privacy. Researchers compare the work of their technology with reading on the lips – this technique also does not recognize all the words, but helps to understand the essence of the conversation, especially when considering the context.
It is important that this technology is concerned about possible threats to privacy, as attackers can use it for remote listening to personal conversations without the user’s knowledge. The researchers themselves emphasize that their work aims to warn society about such risks and promote the development of means of protection against this type of espionage.
The study is supported by the US National Scientific Fund, and the team plans to continue working on the protection of personal information from the latest technological threats.