EU AI Act Transparency Disclosure

Last Updated: August 2026

1. AI Generation Declaration

In compliance with Article 50 of the European Union Artificial Intelligence Act (in force as of August 2, 2026), we clearly disclose that all audio produced by Timbra is generated by Artificial Intelligence.

2. Nature of the AI System

Timbra uses advanced generative models (including parsing, text-to-speech, and audio alignment models) to transform text documents into spoken audio and to generate ambient soundscapes (such as the Reading Capsule). While our models are highly accurate, they are machine-generated and not produced by human voice actors or musicians.

3. Watermarking and Provenance

To ensure transparency and prevent the spread of undisclosed synthetic media, Timbra applies robust tracing mechanisms to every audio file it generates:

• Inaudible Watermarking: An imperceptible audio watermark (using AudioSeal technology) is embedded into the audio stream. This watermark survives compression, cropping, and speed changes, allowing detection systems to identify the audio as AI-generated.
• C2PA Metadata: We embed Coalition for Content Provenance and Authenticity (C2PA) cryptographic manifests into our output files, detailing the origin and generation process of the media.

4. Limitations and Hallucination Risk

Users should be aware of the inherent limitations of AI systems. During the text parsing and chapter generation process, the AI may occasionally misinterpret layout elements, skip text, or mispronounce uncommon words. We provide a “Quality Card” with a Word Error Rate (WER) estimate so you can gauge the fidelity of the parse, and you retain the ability to manually edit chapters before synthesis.

5. Support and Feedback

If you encounter harmful outputs, severe inaccuracies, or have questions regarding our AI transparency measures, please contact our support team at support@timbra.life.