#345: Audiobooks

Visit on Github.

Opened Feb 15, 2019

こんにちはTAG!

I'm requesting a TAG review of:

  • Name: Audiobooks
  • Specification URL: derivative of WPUB
  • Explainer, Requirements Doc, or Example code: Audio Explainer
  • Tests: to come
  • Primary contacts: @wareid, @mattgarrish

Further details (optional):

You should also know that this is this first module of Web Publications Specification (see related issue #344). The PWG recently changed direction a bit as outlined in this blog post. The Publishing WG has not yet drafted specific language for this module. We are working to plan meetings with major audiobook distributors in the USA in March 2019 to obtain their feedback about our proposal.

We'd prefer the TAG provide feedback as (please select one):

  • open issues in our Github repo for each point of feedback
  • open a single issue in our Github repo for the entire review
  • leave review feedback as a comment in this issue and @-notify [github usernames]

cc @iherman @GarthConboy

Discussions

Comment by @markcellus Feb 15, 2019 (See Github)

Looks like the Explainer link needs updating

Comment by @TzviyaSiegman Feb 15, 2019 (See Github)

Thanks @mkay581. I fixed it.

Comment by @TzviyaSiegman Feb 15, 2019 (See Github)

for those reading comments on email the Audiobooks explainer is at https://github.com/w3c/wpub/blob/master/explainers/audio-explainer.md

Discussed Mar 1, 2019 (See Github)

hober: like that it's simple

hober: wanted more from explainer; seems to be missing relevant technology such as podcasts.

hober: audiobooks get very long & large, using a zip file seems maybe suboptimal, the audio is already compressed..?

hober: I assume they have thought of all these things, but I don't see documentation in the explainer of why they were not used

dbaron: did you see the use cases doc for web publications? It seems relevant

hober: I was reminder of torgo's doc on ethical technology principles - having a living document on priority of contituencies. This seems like a case of don't reinvent the wheel - the explainer should make a case as to why this isn't doing that.

cynthia: Right, none of the considered alternatives explain why those were decided against

hober: agreed; it's missing both obvious alternatives and rationales

dbaron: not sure what ePub had for audiobooks in the past.

torgo: do we want them to think about web packaging? We've talked to the ePub people about that before. There's some kind of workshop coming up on this. Is there an obvious, more "webby" approach not listed in the considered alternative that we could encourage them to use?

dbaron: seems possible. I made some notes on the web publications review - audiobooks is the first "sub-spec" of web publications. A bunch of the issues here are related to packaging and how packaging relates to origins and URLs, there were some interesting things in the requirements doc but I haven't gotten to look at how they address those in the spec yet.

torgo: Should we provide this feedback and then bump this two weeks?

hober: Sure, I can try and capture my thoughts on the issue.

dbaron: I won't be here then

torgo: Let's do one week then.

Comment by @cynthia Mar 26, 2019 (See Github)

Out of plain curiosity, why is this restricted specifically to audiobooks? It feels like taking a manifest and doing a regexp like s/mp3/mp4/g would make this a video playlist(?) format, such as what you would get through DVD/Bluray chapters.

Comment by @iherman Mar 26, 2019 (See Github)

@cynthia absolutely correct. The "core" document (wpub) should be the basis of different publication types, which may include audiobooks, good-old-digital-books, scholarly publications, or indeed video playlists. The various publication types (for the time being we use the term 'profiles') are build on top of this core, possibly adding terms and possible requirements that may be profile specific. It is the intention of the group to push any manifest term that may be usable in general into this core.

The reason audiobook is the focus right now is purely pragmatic: there is a dynamically growing market with a very messy practical situation right now. Creators of audiobook players have expressed their frustration with the current status, and a standard is badly needed. Hence the choice of audiobooks as a first incarnation of a specific profile. The WG intends to work (albeit not yet on a rec-track) on other profiles, too.

Discussed Apr 1, 2019 (See Github)

Sangwan: I've looked into this. It's a playlist format, with a bit more. An M3u file plus more. Specificially type 2 audio, while it seems liek it would be related to the ISO BFF web resource packaging would be interested in as well. There is no synchronisation, and there is no 'web' in here. So... it feels like this is a very specific fetaure, but it could fit in to a lot more use cases that they don't seem to be interested in.

I commented: if you do a regex from MP3 to MP4, you've basically just defined the DVD chapter format. A bit weird.

And just no web stuff in here.

Hadley: How did it come to us?

Sangwhan: they used the manifest format they defined for web publications. Subset of publishing... the manifest can have arbitrary resources, including web content. Audiobook part is publishing packages, but only for mp3 files.

I feel like this is a bit short-sighted.

Dan: Is there anything un-web here? Different identifiers besides URLs, etc? HOw is this different to RSS?

Sangwhan: It's JSON instead of XML

Dan: I@m not sure that the criticism that it's not webby... Is there a more webby approach that we can suggfest they take?

Sangwhan: Not criticising that, just a basic remark. There is the entry page which is hte bare minimum. Looks like this would have to be repeated for video, and other formats too. That concerns me.

Hadley: So you'd rather it be done in a generaliseable way?

Sangwhan: Yeah, the publications part is general, and the audio files is a specific use case. But video will have to reinvent their own thing, which is currently ISOBFF, and I'm not convinced that's the best way to do this.

So I'd like to see these two efforts merged.

Hadley: Makes sense to me.

Dan: It does seem like... in their considered alternatives, they have this one thing about ePub 3 audiobooks, and they don't explain why they're not doing that.

Sangwhan: Agreed. I'll add that to the issue.

Dan: Do we need a really specific... It's great to have a teasing-out-the-requirements-of-audiobooks... but do we need something specific to them?

Sangwhan: this is a subset of functionality that they're doing.

Dan: I'm going to be in really cold Canada next week, and I might try to chat to Tsviya... and get a bit more context.

Sangwhan: This is a valid use case for video, so I'd like to explore that.

Kenneth: Who would be using this? Apps like netflix for audiobooks? Audible?

Sangwhan: Basically Americans. I think they use their own thing, (probably some variant of) the pocket mobi format.

Dan: What is the use of audiobooks worldwide?

Sangwhan: They're more common in America because of the drives.

Peter: agreed

Kenneth: and then they need HTML at that point?

Sangwhan: You'd use HTML to push a button to say "play chapter 2"

...I will write feedback; Dan you talk to Tsviya -- and let's talk after that.

Dan: Comments from mkwst. David (who is not here today) had some comments. Does anyone have feedback? Or should we wait?

(Hears silence)

Alice: Looks like mkwst had some plans to do work on the explainer, might take some time.

Dan: Any time sensitivity?

Alice: Probably not.

...Punt for two weeks (when David is back)

JUMPBACK:

Lukasz: I will write feedback about this on the issue.

Dan: Might be useful to mention the use cases. With the new restrictions of 3p cookies on the platform, there are considerations that should be taken when designing features like this.

GOTO PORTALS;

Comment by @torgo Apr 3, 2019 (See Github)

@TzviyaSiegman can we discuss a bit at the AC meeting next week?

Comment by @TzviyaSiegman Apr 3, 2019 (See Github)

@torgo more than happy to discuss!

Comment by @torgo Apr 17, 2019 (See Github)

@TzviyaSiegman we didn't get to discuss this at the A.C. unfortunately - let's find a time to talk sometime in the coming week.

Comment by @wareid Apr 17, 2019 (See Github)

@torgo @TzviyaSiegman is on vacation (a well-deserved one) for a few weeks, I'm happy to discuss anything with you if you want to this week, or we can wait until she returns!

Discussed May 1, 2019 (See Github)

Tess: last i remember we were pending external feedback from tsvia

Dan: i am due to talk to her tomorrow about web publications review

Tess: in addition to punting this until we hear back from T, there was conversaiton on audiobooks

Discussed May 1, 2019 (See Github)

Dan: also 345.

Dan: I had a call with Tsviya about both of these a couple of days ago. They're working on our feedback. In 344 there has been some back-and-forth, some of which was orthogonal to the main review. Tzviya noted that; the ball is in their court because they're looking to update the spec, in particular to address the issues David raised. We should probably revisit at the f2f.

Dan: Milestone update.

Dan: Regarding audiobooks, they're going to be updating the explainer.

Tess: Most of my feedback was questions that could have been answered in the explainer.

Dan: Some of the decisions they've taken in the spec are driven by specific stakeholder needs, and this should be noted in the explainer. Trying to balance concerns between web community and publishing community. Hoping they have an updated explainer we can take a look at at the f2f; ball is in their court, issue is pending external feedback.

Dan: Tzviya is looking for feedback also on how to make TPAC more welcoming to new people.

Comment by @cynthia May 21, 2019 (See Github)

@hober and I discussed this during the Iceland F2F - we've been told that there is a updated explainer coming up, and we'd be glad to review this again when that is in place.

Comment by @rektide Jul 22, 2019 (See Github)

Really hoping to see https://github.com/w3ctag/design-reviews/issues/354 advance, reading this part of the spec on Web Packaging:

The Working Group is very interested in Web Packaging as a future option for Audiobooks (and other web publications), but due to timing, we have decided to proceed with a temporary option in the Lightweight Packaging Format (https://w3c.github.io/pwpub/), until there is a viable web option.

This audiobook spec includes a manifest.json, which among other things enumerates all the files & includes some metadata. This is a good thing for the future, because I don't believe there's any API as a part of Web Package or otherwise that a page can use to discover other bundled content.

In a webpackaging world, I feel like some adjustment to the manifest would be required. If the manifest refers to filenames, is it up to the browsing context to figure out what URL the manifest was at, and to join the paths together? Alternatively, the manifest could include canonicallized full urls, rather than filenames or paths, as these would be how the resources identify themselves inside the bundle.

Venturing further afield, the web's inability to observe available resources is also present in fetch, which won't tell you that, if you fetch a resource, that other resources are being PUSHed in reply to that fetch ( whatwg/fetch#65). In general, approaches to "audiobook" might be different if, when a bundle is loaded, or resources are pushed at the page, the page had some means to detect those resources which it now "has".

Comment by @hober Dec 5, 2019 (See Github)

Hi,

@dbaron, @plinss, and myself took another look at this during our Cupertino F2F. All of our previous feedback in w3c/audiobooks#9 has been duly considered, and design review issue #423 captures the remaining work on our side. We're therefore closing this. Please request a new review if the design substantially changes. Thanks!

Comment by @wareid Dec 6, 2019 (See Github)

Thank you!