ZedAI or EPUB3 as a master format?

So, ZedAI is DAISYs proposed master format in a single-source publishing workflow. But as far as I can tell, you can't describe an audio book with ZedAI. So for narrated audio books (full-text or not), I fail to see how it can be used as the master format.

However, EPUB3 can be used as a master format for narrated audio books. So then I wonder, why not use EPUB3 as a master format for text-only books as well? What is there to gain from choosing ZedAI over EPUB3?

Is there any blog-posts or articles comparing EPUB3 and ZedAI? What are the advantages and disadvantages of choosing one or the other as a master format? In what situation would one choose one or the other?

ZedAI is an xml (text) master format, not an audio one, so that omission was always by design. Part B of the specification would have been to define a new DAISY audio format, but that work was eventually abandoned when EPUB 3 came along and it became clear that the IDPF would fully embrace accessibility.

If you can make do with HTML for text authoring, or need a media overlay and audio with your text master, then you can certainly use EPUB 3. Don't forget that many of the people involved in ZedAI were also involved in EPUB 3's development (notably Markus Gylling), and the goal was to ensure it could be as expressive as possible for text and audio.

Where ZedAI has advantages, though, is in the ability to easily extend to define granular markup to handle specific problem domains, like bills and poetry and cookbooks and plays (I've done some experimental development already on the latter three outside of the existing framework and core modules). HTML's limited tag set makes coordinating the authoring of markup for complex formats difficult -- between authors and from even from book to book for the same author -- and often heavily depends on CSS or layering of multiple semantics to express what can be more easily done (and understood) with a single tag. To get the full richness of the Z39.98 structural semantics vocab in epub 3, you'd also need to import it and prefix all the terms, which is just another authoring nuisance.

ZedAI was intended to to provide a component-based architecture that anyone in the DAISY consortium could use to build new formats, and be able to effectively interchange with other organizations. It was also designed to lower the burden of implementation so that each organization doesn't have to go through the potentially painstaking process of developing HTML tagging and semantic structuring rules for each format type, only to discover that each of these formats has to be reworked on exchange. It also allows easy prototyping of new grammars where you may have unique needs (e.g., throw-away print braille ingestion files).

When it comes to evaluating which markup is going to work best for you, you're ultimately going to have to weigh what works best for your organization's needs. I wouldn't say that one approach is definitively better than the other, they each just have different strengths.

Thanks for the detailed answer! It will be very helpful when we evaluate how we want our production line to look like in the future.

Regards
Jostein / NLB