Evixar identifies fake videos generated by AI using acoustic watermarks to prevent abuse, developes content protection service and starts accepting trial programs, 26 December 2023

Evixar has developed a content protection service called EAF (Evixar Audio Forensics) that uses a special sound embedded with encrypted data (watermark sound) and fingerprinting technology for sound sources stored in a database, and has begun recruiting participants for a trial program. It is scheduled to start offering in January 2024. This will prevent the misuse of deepfake (human image synthesis) videos generated by AI (artificial intelligence). Promoting the development of technology that can identify deepfakes and creating international rules for checking the safety of new services before they are released are major topics on the agenda of the Group of Seven developed countries (G7). Evixar's acoustic communication technology has been patented in Japan, the United States, China, and South Korea, and we will cooperate with the Japanese government to strengthen content protection. In principle, the trial program is provided free of charge, and the price for the commercial stage has not yet been determined.

Flow of content registration process by EAF (Evixar Audio Forensics)

Flow of EAF (Evixar Audio Forensics) for false video detection

EAF（Evixar Audio Forensics）

"EAF" allows users to register and log in to the system on the web, upload video data, and the system will embed a watermark in the audio of the video and generate acoustic fingerprint data from the audio of the video. The watermarked material can then be downloaded and distributed via various media. On the other hand, when a suspicious video becomes a topic of conversation on a social networking service (SNS), the video is downloaded and uploaded to the system for watermark detection and fingerprint match search. As a result, if no watermark is detected and no fingerprint matches are found, the video can be judged to be generated AI. Applications for the trial program will be accepted on a first-come, first-served basis using the following form at any time.

「EAF（Evixar Audio Forensics）」 - Beta Program Pre-Application Form

https://product.evixar.com/digital-forensics

Evixar has developed an acoustic communication technology that automatically recognizes watermarked sound and fingerprints by devices such as smartphones, smart glasses, and robot terminals, and controls the devices in real time. "HELLO! MOVIE", a free application that allows users to enjoy movie subtitles and audio guidance via smartphones and smart glasses, and "subtitle glasses" loaned by movie theaters to enable barrier-free screenings. We have obtained patents in Japan, the U.S., and China for audio and subtitle guides for movie theaters, and in Japan, the U.S., and South Korea for subtitle guides for theaters and performing arts, respectively. They are also used in the field of disaster prevention, such as acoustic communication from a drone to a smartphone to provide disaster prevention information in places where people cannot go, and in non-contact multilingual services in museums.

The main feature of these applications is that they can be controlled without modifying the original content (in the case of a movie, the full-length movie), and we have been studying the use of these technologies for deep faking countermeasures. In the U.S., a presidential decree has already started requiring the embedding of digital watermarks in generated AI, and Evixar will contribute to preventing the misuse of deep-fake videos.