ZedDist EPUB Contributions
From zedwiki
This page provides an overview of given and possible technical interrelations between Z39.86-2011 and EPUB 2.1/3.0.
Contents |
OPS-related
Replace the EPUB 2.0 DTBook support with the Z39.86-2011 equivalent
EPUB 2.0 supports the DTBook 2005-2 DTD as an alternative to XHTML for textual content. In Z39.86-2011, DAISY is replacing DTBook as a distribution format with a schema that is a derivative of XHTML5.
This is perceived as a win-win situation for all, assuming that the main EPUB text content format also moves in this direction. The DTBook replacement contributed by DAISY would be a strictened subset instead of a completely different grammar as in EPUB 2.0. This reduces the costs of implementing support for both schemas significantly, and makes the user experience more homogenous.
Mechanism for structural semantic extensions to XHTML5
We wont be providing some readybaked solution here, but develop one jointly with the epub WG.
Potential solutions include:
- namespace-based extensions to XHTML5
- role attribute (used beyond HTML5's definition of it), or a separate attribute (<a z:type="noteref" href=""/>)
- external semantic sheets
(Further outline in ZedDist_Document_Text)
This table lists dtbook concepts that can not be translated directly into HTML5.
| Concept | Description | Implementation |
|---|---|---|
| front- body- and rearmatter | Major structural divisions in a document. These are elements without any visual representation, but are still important for reading software and other user agents | A block element that can contain other block markup, including structural headings. The HTML5 element "Section" with a role or class attribute describing the element would be a good candidate |
| Notes/Annotations | Foootnotes, endnotes and annotations in the text. These are elements that are referenced from other parts of the document. Elements of this type is a container for other block mark-up (the actual note content) and in order to be referenced an ID attribute is required. Though book publishing makes a distinction between footnotes, endnotes and annotations, they are not different from a mark-up point of view. | A block element that can contain other markup: <aside> or <div> with a role or class attribute with the value="note" or "annotation"
|
| Noterefs/Annorefs | Elements that reference notes and annotations - or more precise, reference an ID in the document. The reference must be a single text element, contained in either a block element or in an inline element. | The notereference is the text element of an "a" element with a href attribute with the value of the referenced note/annotation |
| Production note (prodnote) | Contains language added to the alternative-format version by the producer. A prodnote is a container for other block markup. A common usage is the description of images or other visual representations in the text. It should be possible for the user to "turn of" prodnotes in the text. | A block element that can contain other block markup - e.g. div |
| Pagination | Navigation points in the text representing pagenumbers. Note that the actual page is not represented as a container, though a virtual page container can be constructed from the pagenumbers. Pagenumbers are inline elements with attributes that describe their properties. It is important for a user agent to know if a pagenumber is in roman, arabic or any other format and the ID of the pagenumber is important for navigation purposes. | In HTML pagenumbers wil be contained in a span element with attributes that describe its properties
|
| Headings, other than <h1>-<h6> | Headings that are not part of the main structure of the document, e.g. headings in lists and <div>. The main structure of a document is defined by sections and headings, but it is often usefull (and necessary) to insert headings in mark-up that is not part of the main structure, e.g. in list and divs. | HTML5 does not have an element of this type. It is possible to work around the problem, but an element with no structural value (unlike h1-h6) would be preferred. |
| Imggroup | A container for one or more images and their captions and/or descriptions. Especially in more complicated textbooks it is not unusual to have several captions referencing the same image and vice versa: a number of images can be referenced by one caption. In order to do this the caption must be able to reference the ID of the image. | At the moment HTML5 does not have elements that mimic this functionallity completely. |
| Semantic annotation attribute | An attribute that allows semantic annotation of any element | This attribute goes beyond the ARIA-role attribute, to the extend that it acts like a placeholder for the actual element defined by the DAISY Consortium. RDFa 1.1 could provide a workable solution otherwise we will probably create our own solution |
Integrate support for speech synthesis pronunciation instructions
- An increasing number of reading devices ship with built-in TTS
- For the print disabled in particular, this is a crucial feature
- But good (read: low rate of mispronunciation, near-correct prosody (stress, intonation) is a general device usability concern
Problems
The use of TTS within the DAISY realm over the past decade has identified the following core problems:
- While TTS quality has improved, the quality varies dramatically between languages (the bigger the language, the larger the market for TTS vendors)
- For academic and specific-domain content, most TTS's fail miserably (too many terms not covered by hardwired lexicon). For professionals and students, this means that consuming content using TTS is painful at best and impossible at worst.
Requirements
- A way to link to publication and/or document-wide pronunciation lexica
- A way to specify atomic pronunciation and prosody rules inline (lexicon-provided generalizations are not sufficient)
Solutions
Existing languages that we can reuse:
SSML
- PLS defines a lexicon document type for pronunciation instructions
- SSML 1.1 defines a set of elements that can be adapted for inline usage. (see also Z3986-AI which defines an attribute-axis-only adaption for SSML-based pronunciation instructions).
... note: SSML is partially or fully supported by TTS engines such as cepstral, AT&T, Loquendo, neospeech, and more.
CSS3
http://www.w3.org/TR/css3-speech/
NCX-related
Replace the EPUB 2.0 NCX support with the Z39.86-2011 equivalent
Z39.86-2011 makes a number of changes to the NCX, all of which have positive impact on EPUB:
- The NCX schema is being modularized so that customized NCX schemas can be built for particular application domains
- Increasing the terseness of the syntax in order to reduce memory consumption for non-streaming applications
- Adding a method to add structural semantic metadata to NCX links and groups of links
OCF-related
Container Level Fallback
The EPUB 2.0 spec describes how to offer fallback descriptions for individual media files, using the OPF file (http://www.idpf.org/2007/opf/opf2.0/download/#Section2.3.1), but not how to offer fallback at the level of the entire document.
If EPUB really wants to be a format for richer multimedia across a spectrum of devices, a container level fallback mechanism would be a useful way to allow publishers and device makers to make transparent use of a single EPUB distribution. Even without "profiles" in EPUB, there might be a case for wanting to have a package with multiple varieties of content:
- multimedia content for a high end experience
- regular HTML content as a default
- text-only content or audio-only content as a basic reading experience.
Being able to be explicit about the precedence ordering of these would be tremendously useful.
Streaming-Friendly Container
The EPUB 2.0 spec allows two physical container types, filesystem and ZIP format. Extending that to support a TAR/GZip format would provide a single distribution unit and a format that is more conducive to streaming content. Having file information at the start of the stream makes it possible to navigate the stream by offsets, making it possible to deliver EPUB content from a central server.
Multimedia-related
Full-presentation multimedia synchronization
- Markus/Marisa: Full-presentation multimedia synchronization
Accessibility of media islands (structural navigation)
Other Items
Dear to us, but not particularly related to DAISY/accessibility:
- interpublication linking
- shareable/interoperable annotations/bookmarks (Compare Z38.86-2005 Bookmark DTD)
- ... more?
