How Embedding Differs by Format

C2PA uses container-native embedding methods for each format type. The manifest data is placed in the format-standard location for extension data, ensuring compatibility with existing tools that process each format.

Images

13 image formats with JUMBF manifest embedding

Audio

6 audio formats across RIFF, ID3, and ISO BMFF containers

Video

4 video formats with ISO BMFF and RIFF embedding

Documents

5 document formats with custom JUMBF/COSE embedding

Fonts

3 font formats with custom JUMBF/COSE embedding

Text Content

Unstructured text - articles, social posts, email, and any plain text - is covered by C2PA Section A.7, which Encypher contributed to the specification. Text does not have a file container, so provenance is embedded using invisible Unicode characters within the text itself.

Text is not included in the MIME type table because it does not have a single MIME type - it covers text/plain, text/html content bodies, and any unstructured text context. See Text Content Provenance and C2PA Section A.7 for details.

Verification Pipeline

Encypher uses two verification pipelines depending on format:

• c2pa-python-native

  Formats natively supported by the c2pa-python library: all 13 image formats, 4 of 6 audio formats (WAV, MP3, M4A, AAC, MPA), all 4 video formats. The c2pa-python library handles manifest extraction and signature verification directly.

• custom-jumbf-cose

  Formats requiring custom implementation: FLAC, JPEG XL, all 5 document formats, all 3 font formats. Encypher implements JUMBF embedding and COSE signing directly for these formats. Verification uses the same cryptographic primitives but with custom manifest extraction.

Signing via the API

All 31 MIME types are accessible through a single signing endpoint. Format detection is automatic based on file content, not file extension: