How TransparentMeta works

TransparentMeta is a Python library that integrates seamlessly into your application. Once you import TransparentMeta, with just a few lines of code, you can add transparency features quickly without disrupting existing development workflows.

Writing

When AI-generated audio is produced, you can call a TransparentMeta routine to write cryptographically signed metadata directly into your audio files. This metadata includes key details such as generation timestamp, model identity, and content origin.

Reading

TransparentMeta provides tools to read and verify embedded metadata from audio files at any stage of the distribution chain. This functionality allows platforms, regulators, or end-users to confirm whether audio content is AI-generated and ensures the transparency information remains intact. The verification process validates cryptographic signatures to confirm data integrity and detect any unauthorized modifications.

Reading is done by calling a simple TransparentMeta routine, making it easy to integrate into existing systems.

Use case

The typical use case for TransparentMeta involves a generative audio AI company that wants to label its AI-generated audio content with metadata that flag it as AI-generated. This is crucial for compliance with AI transparency legislation, such as the EU AI Act and the California AI Transparency Act.

The worklow usually looks like this:

  1. The company generates audio content using its AI models.

  2. Just before distributing the audio, it uses TransparentMeta to embed cryptographically signed metadata into the audio files.

  3. The audio is stored and distributed through various channels.

  4. When the audio is accessed, platforms or end-users can use TransparentMeta to read and verify the metadata, confirming the audio’s AI-generated nature and its origin.