Addition of "EPUB 3 Text-to-Speech Enhancements 1.0"

[murata2makoto]

I think that EPUB 3 Text-to-Speech Enhancements 1.0 should be added in 3.2.

[iherman]

@murata2makoto, there are currently 7 active WG notes under the "control" of the EPUB WG, and one more under the "control" of the Publishing WG. There is no reason to make an exhaustive list of those in the charter; the introductory text says (emphasis is mine):

Other non-normative documents may be created and/or maintained such as:

I have picked the one from the Audiobooks and, somewhat arbitrarily, one from the EPUB related list (probably the one which has the highest interest in practice). I would prefer to stop there, not to overcomplicated the charter text.

(Nothing against TTS, of course.)

[murata2makoto]

@iherman

This feature is used by Japanese textbook publishers but is nevertheless dropped from EPUB 3.3 for the lack of world-wide implementations. I strongly want to have this spec on the list.

[iherman]

@iherman

This feature is used by Japanese textbook publishers but is nevertheless dropped from EPUB 3.3 for the lack of world-wide implementations. I strongly want to have this spec on the list.

I presume you mean 3.2, which is a list for "maintaining notes". You seem to argue that you would want it as part of the Rec, which is not a statement that should be done in the charter. Note that the WG decision is that substantial changes on the EPUB 3.3 standard would have to go through a complete FPWD->WG draft->CR->PR->Rec route, which may not be under the purview of this Working Group anyway (i.e., a new charter would be necessary).

Bottom line: I do not know what you intend to achieve by adding a new item to this list (I am obviously concerned that we will end up listing all the 7+1 note references, which I think will just complicate the charter review.)

[murata2makoto]

I am not arguing for incorporating this note as part of 3.4 (at least now). I would like the charter to clearly allow possible maintenance of this note especially when APA finalizes "Specification for Spoken Presentation in HTML". #13

[iherman]

I would like the charter to clearly allow possible maintenance of this note especially when APA finalizes "Specification for Spoken Presentation in HTML".

But that is exactly what the introductory paragraph says, which refers to all notes published by the earlier Working Groups, with this note being part of the "such as"...

I am sorry, @murata2makoto we seem to mutually misunderstand one another here. Again, I do not know what you intend to achieve (and which is not already in the text) by adding this item to the list.

[murata2makoto]

@iherman

Then, why does 3.2 reference notes? Some are very important but not all of them, as I see it.

[iherman]

Then, why does 3.2 reference notes? Some are very important but not all of them, as I see it.

First of all, it refers to two notes only + the documents on testing (EPUB and audiobooks, respectively; testing maintenance is also referred to in the charter). LPF is the only note published with by the audiobooks wg, that is why it is important to get it there. As a matter of equity between the two originating working groups, this leads to pick one note out of the ones listed on the EPUB WG's list. Maybe I am wrong, but I regard the A11y techniques' document is probably the most widely used and most important note among those. That is how we got to what we have.

Aren't we bike shedding at this point?

[murata2makoto]

Aren't we bike shedding at this point?

The use of SSML and PLS is crucial for Japanese textbooks (unless the voice is included). Although I certainly think that the A11Y techniques document is important, I would argue that this note is more important for Japanese textbooks.

[shiestyle]

@murata2makoto Japanese textbooks is one of the categories in publishing industry, even in Japan.

Although TTS in Japan will be more important after EPUB 3.3 is published, I don’t think it is necessary to pick up as an example in this charter.

I agree with Ivan’s explanation.

[mattgarrish]

This sounds like an issue of exposing all the work of the various groups better. We've talked about having a page that lists all the REC track docs, WG Notes, CG Final Reports, etc. that the various publishing groups have created so you don't have to go hunting for them. I'm all for that, but a charter document isn't a helpful place to have it.

[murata2makoto]

Japanese textbooks is one of the categories in publishing industry, even in Japan.

True. It is also true that no trade book publishers in Japan create textbooks and no textbook publishers in Japan create trade books. This is special in Japan.

[murata2makoto]

This sounds like an issue of exposing all the work of the various groups better. We've talked about having a page that lists all the REC track docs, WG Notes, CG Final Reports, etc. that the various publishing groups have created so you don't have to go hunting for them. I'm all for that, but a charter document isn't a helpful place to have it.

Matt, JDC is involved in creating DAISY textbooks in Japan. Since TTS is so error-prone in Japan, DAISY textbooks have recorded voice always. Creating such DAISY textbooks is a significant burden to volunteers. Meanwhile, textbook publishers in Japan started to use PLS and SSML recently.

I have always thought that Publishing@W3C does not understand the current situation in Japanese textbook publishing well and pays too little attention to PLS/SSML. It is not merely one of the many specifications. It is much more important than that.

[shiestyle]

@murata2makoto Even if TTS is important in Japan, it will not be the most important issue in accessibility area for publishing in global.

At least in Japan, TTS is a small part of issues in publishing industry, and your claim doesn’t seem to make sense.

[murata2makoto]

@shiestyle

s/publishing industry/tradebook publishing industry/

[shiestyle]

@murata2makoto Textbook publishers are part of publishers in Japan, right? Even if TTS for textbook publishers is more important than for tradebook publishers, TTS will not be the most important issue for all publishers in Japan and also in global, because the market and readers who need it are not large.

If we will follow your logic (important in a specific area is important globally), we cannot give priorities properly. You should show the reason why TTS is more important than EPUB-A11Y Techniques and the critical influence when TTS is not listed in the charter.

[murata2makoto]

Textbook publishers are part of publishers in Japan, right?

This list of board members of "Digital Publishers Federation of Japan" has no textbook publishers. Meanwhile, the list of board members of "Textbook Publishing Association of Japan" has no trade book publishers.

TTS will not be the most important issue for all publishers in Japan and also in global, because the market and readers who need it are not large.

I understand that this is your position. Is this also the position of your company or Japanese trade book publishers in general?

[shiestyle]

@murata2makoto I just sent what you should answer and don’t reply questions to questions.

Anyway, it’s not my position but a fact that readers who need TTS is smaller than who don’t need.

And a specific industry group without textbook publishers will not define that textbook publishers are not publishers in Japan. It seems a common sense.

[murata2makoto]

You should show the reason why TTS is more important than EPUB-A11Y Techniques and the critical influence

I am sorry for my delay. It is not easy to respond to this question in a concise manner.

First, the year before last, the Japanese National Diet Library conducted a thorough survey and published a report about accessible e-book services in libraries. In particular, it shows which type of print disabilities requires which accessibility feature. Based on this study, I estimated which feature is needed by how many people. Features specific to Japan are ruby, space-separated phrases (wakati-gaki), detailed reading mode, and switching of vertical or horizontal writing direction. Each of them is required by more than 1.5 million people. But none of these features are sufficiently covered by WCAG, EPUB A11Y, or EPUB A11Y Techniques.

Second, it is generally agreed that TTS for Japanese text makes many mistakes. I am aware of one government report and one book that strongly advocate the use of SSML for eliminating such mistakes. I also know that the biggest textbook publisher in Japan uses SSML very heavily and that a textbook e-reader adopts PML. On the other hand, I also know that some people think that SSML is a red herring. They think that users can live with TTS errors and further think that other features such as navigation is more important than making TTS error-free. (But such guys are typically blind geniuses!) It is difficult to predict the future, but I think that the semi-automatic addition of SSML or PLS by natural language processing is a promissing candidate. Again, as far as I know, none of these issues are sufficiently covered by WCAG, EPUB A11Y, or EPUB A11Y Techniques.

I am not sure if WCAG, EPUB A11Y, or EPUB A11Y Techniques should try to cover all accessibility features specific to some natural language. Ideally, just like JLreq and other xLreq specifications developed by several subgroups of the I18N WG, there should be "Requirements for Accessibility Features Specific to the Japanese Language" and so forth. But we are very far away from such an ideal world yet.

[shiestyle]

@murata2makoto Thank you for your reply.

But it’s not the answer we want, we don’t want to hear how TTS is important in Japan. As you may not understand, I also think TTS is one of the most important issue in coming several years for publishers in Japan.

But such situation couldn’t explain that TTS is more important than metadata for A11Y or conformance to WCAG. I know TTS for Japanese has some quality issues but W3C note for TTS will not solve them.

If you think TTS should be included in EPUB-A11Y Techniques, discuss at A11Y-TF or some other places. This charter is not appropriate place for that.

[murata2makoto]

@shiestyle

The content of the note should have been a part of 3.3 as it is crucial for TTS. Updating the note in sync with "EPUB 3 Text-to-Speech Enhancements 1.0" is the least the WG should do. In this sense, it is more important than EPUB A11Y Techniques.

I do not believe that you think TTS is important, because TTS without SSML/PLS cannot be very good for the Japanese language and also because you have said and written something like "TTS will not be the most important issue for all publishers in Japan and also in global, because the market and readers who need it are not large."

[shiestyle]

@murata2makoto I’m sorry but I cannot think we will understand mutually even if we continue to discuss here.

TTS is not mandatory for EPUB-A11Y and then you should discuss in accessibility area if you think it is more important than or as same as A11Y metadata or WCAG. It should be discussed as the future versions, of course.

So, I propose you to gather who has the same recognition you stated below.

In this sense, it is more important than EPUB A11Y Techniques.

[iherman]

Interesting as it is, this is not a discussion here and now... it is typically the discussion that should happen, at some point, at a WG level. We are just discussing the charter here, and the tiny issue of which document reference we use in the charter as an example.

I would propose to close this issue, discuss this problem somewhere else, and/or gather the arguments pro and con for a future discussion at the maintenance working group (which, remember, does not yet exist!)

[murata2makoto]

@iherman

No, I cannot live with your proposal. I will contact more people and encourage them to participate in this discussion and even think about formal objections.

[dauwhe]

@iherman does it do any concrete harm to add one more example? We are not committing to work on any particular note.

[shiestyle]

@dauwhe It was a tiny work to add one more example but after these discussion we should not do so now because following Makoto's request will indicate that we will accept a kind of inconsistency of EPUB 3.3 and EPUB-A11Y 1.1 and we should not behave without consensus logically.

[iherman]

I propose something a bit more, to avoid such discussions. I will put in a PR that lists all current WG Notes from the two WGs. In alphabetical order, to avoid any sort of intended or perceived bias.

PR is coming soon.

[iherman]

See PR #15; this should make everyone happy... :-)