Collaboration Between Data Analytics Lab and Evixar to Combat Misinformation on the Internet

Joint Research to Combat Misinformation

In a proactive move against the rising tide of misinformation and disinformation on the internet, Data Analytics Lab Inc. and Evixar Inc. have initiated a collaborative research project that leverages advanced audio signal processing technology. This partnership aims to enhance techniques for verifying the authenticity of online audio-visual content, driven by the urgent need for robust solutions in an era dominated by social media and AI-generated materials.

Background and Objectives of the Research

The rapid proliferation of social networking services and AI technologies has exacerbated the risks associated with deceptive information online. The emergence of deepfakes and manipulated media poses significant challenges not only to individual users but also to societal trust as a whole. In response, there exists a pressing demand for techniques that can guarantee the authenticity of audio and video content. This joint research project is part of Evixar's involvement in the Ministry of Internal Affairs and Communications' initiative on developing technologies to counter misinformation online.

The research will focus primarily on Evixar’s Evixar Audio Forensics (EAF) — a cloud-based system designed to combat misinformation using audio watermarking and fingerprinting techniques. By synergizing Evixar's methods with Data Analytics Lab's expertise in AI and data analysis, the following research themes will be explored:

1. Development and systematization of test data utilizing diverse generative model approaches.
2. Enhancement of technologies for discerning the authenticity of audio-visual content.

The outcomes of this research are expected to align with international AI governance trends, standardization efforts (such as those led by organizations like C2PA), and initiatives like the Hiroshima AI Process. The ultimate goal is to contribute to the establishment of a reliable information distribution framework as a fundamental social infrastructure.

Foundations of the Research

The collaboration has its roots in the successful acceptance of Evixar's proposal by the Ministry for two consecutive periods under the project targeted at developing and demonstrating technologies to counter misinformation. The EAF technology is built on the following foundational techniques:

1. Audio Watermarking: This technique embeds provenance information and data (known as electronic watermark) into audio signals in an imperceptible manner. It allows for the detection of alterations or copies of the content, making it challenging to remove since the data is embedded directly within the audio itself, providing enhanced resilience against tampering.
2. Audio Fingerprinting: This technique converts segments of audio data into unique features, enabling identification akin to a fingerprint. By matching these features, it becomes possible to determine whether the content is identical or has undergone modifications or alterations.
3. Passive Authenticity Assessment Using Audio Fingerprinting: This method analyzes audio segments that are suspected to be AI-generated, utilizing audio fingerprinting technology to evaluate their unnatural characteristics, thus indicating the presence of generated content.

Research findings are expected to be reported by March 2026 as part of the Ministry's documentation on technologies against misinformation. This collaborative effort represents a significant advancement in addressing the contemporary challenges posed by false information online.

About Evixar Inc.

Website: Evixar
CEO: Jun Takigawa
Location: Shinkawa 1-17-22 Matsui Building, Chuo-Ku, Tokyo
Established: March 12, 2004
Capital: 100 million JPY
Business Focus: Software development based on audio signal processing and providing acoustic communication solutions.

About Data Analytics Lab Inc.

Website: Data Analytics Lab
CEO: Masahiko Kondo
Location: Shiodome East Side Building, Tsukiji 5-4-18, Chuo-Ku, Tokyo
Established: April 26, 2019
Capital: 20 million JPY
Business Focus: Data analysis, AI application support, AI development, and training services for data personnel.

For inquiries related to this project, please contact:
Data Analytics Lab - PR Contact: [email protected]
Evixar - PR Contact: [email protected]