Meeting minutes
Recommend a format for the evaluation date: Issue #2725 in EPUB specs repo.
<AvneeshSingh> w3c/
mgarrish: required date format for the evaluation date certifiedby etc. right now any date formate is currently supported but we need to standardize we thought we would use ISO date standard
<mgarrish> w3c/
mgarrish: require YMD, YM, or Y this was brought up in the WG call
We will allow any ISO DATE, DATE/TIME display can be customized
… proposed dropping it to a SHOULD because a MUST would need to require a version # bump to 1.2. If we leave it as a SHOULD this would cause more processing to ensure its a localizable DATE field or some other non ISO date which couldn't be localized. ISO8601 format. Not sure in terms of Normative priorities.
AvneeshSingh: EvaluatedDate is not Manditory, but we need to know if you do include a date will it be a MUST or SHOULD .
… ACE currently doesn't check but we can adjust ACE to make it more elevated.
Hadrien: i am in favor of making it a MUST if we bump the version number.
AvneeshSingh: AccessModeSufficient is also in the same boat, currently a MUST but if we get pushback we can make it a SHOULD.
Charles: I also agree this should be a MUST as we will more than likely be a 1.2
mgarrish: Put in an Editors Note stating this we will make this a MUST but will require a version bump to signify folks to take a closer look
mgarrish: I don't think this is a problem, the PR is currently a MUST
… ISO8601 don't care if its just a DATE or DATETIME it won't affect display.
mgarrish: I will add the time option and keep it as a MUST and add an Editors Note about this requiring a version # bump
RESOLUTION: keep the requirement a must with the addition of allowing time values and add a note explaining the potential impact on the version number
Using the UX display guide with Daisy files: Issue #401
<AvneeshSingh> w3c/
AvneeshSingh: using the guide can provide same display for EPUB or DAISY files.
Hadrien: increased number of both DAISY and EPUB files intermixed.
… an interest in provide a11y metadata regardless of format. consistency for UX
George: important need to do. most people with disabilities were getting their books from one place or another and traditionally NLS, Learning Ally - DAISY audio books but those books were always the same. some navigation, no page nav, audio recordings.
… organizations could display metadata based on date produced, but if new stuff could get this metadata, and could retrofit old data to add this metadata.
… not sure if you can go put it back into that old book
Hadrien: if you are using ONX, or OPDS there is a solution, but not sure we have a uniform solution which is why I have the PR.
AvneeshSingh: we have metadata about pagenumbers, audio only etc. but no concept of certification for DAISY. how can we make this work.
George: Internet Archive only has 1 entry for TOC but not really useful.
AvneeshSingh: what is the way forward?
Hadrien: recomment sharing the metadata in another format
AvneeshSingh: DAISY is by default accessible which is why it didn't really need a11y metadata inside DAISY files.
… DAISY pipeline does have a way to inject a11y metadata but the OPF will need to be updated.
Hadrien: I know there are DAISY pipelines with TTS engines, if it is automated. some may prefer Human Narrated or presence of audio is important.
George: another issue Recording for the Blind uses volunteers, 2 hour stints, so you don't get the same narrator for the entire book. No for Human Narrated books you have different characters have different voice actors which is a great experience
Morris: recording of gender of narrators has come up. some people prefer a particular gender, may want a female or male voice. we wanted to de-emphasis gender but this is at odds what is being requested.
AvneeshSingh: DAISY" metadata some kind of crosswalk with completely solve this? what are the next steps?
George: we could list the features in the DAISY file a checklist.
… ABC that checklist could be helpful in determining if you will get this book or not.
AvneeshSingh: we can take this on by DAISY.
Hadrien: I agree this is DAISY specific thing and can help with OPDS.
George: would this apply to audio books.
… BISG is working on an Audio book specification. traditionally mainstream would not have page numbers, image descriptions etc. the checklist idea seems interesting.
AvneeshSingh: things can be inferred by the structure from the NCC or OPF file. there is no programmatic way into the files this has to be in the metadata
Hadrien: if there is an audio book spec. it would warrant covering them in a display guide.
AvneeshSingh: Plan: to have a discussion with DAISY members, I can make a checklist which details we can record from the DAISY publication.
… especially want feedback from Scandinavian members.
Madeleine: define a set base on date
… give them a name then that profile can be applied for a batch.
George: class of books for dyslexics where images were not described. i think called Trade books
Shall we add a note on audio only about type of narration to help users identify AI voices? (Issue #559
<AvneeshSingh> w3c/
AvneeshSingh: need for AI, vs. human vs synthetic
George: human or not
Charles: knowing if it is the famous person or an AI representation
George: yes we did this at CSUN John Wayne speaking a book.
Hadrien: how do I identify a synthetic voice. can i identify the gender, and 3rd might be a cloned voice knowing who it is a clone of is also important to know.
… identify audio book narrated by a human being will be very important as well
AvneeshSingh: Schema.org metadata group need to look into this, APA and BISG to come up with some new metadata.
Hadrien: there is already a w3c audio spec so what is BISG doing?
George: lot of work going on right now. should be standardized. I don't know where that is happening. BISG is not a standards group only applying standards.
AvneeshSingh: MARC, ONIX will also need to be harmonized
Madeleine: we need to coordinate what are the categories are and then we can define the vocabulary
George: I can introduce Madeleine with the BISG folks
… they wanted me to join that audio book group.
AvneeshSingh: is there a subset of folks with metadata folks at BISG.
George: Clair with OCLC had a colleages Morris join call
Morris: I need to clear it with Clair
George: OCLC is also working with MARC records.
Hadrien: pointing out this is also important for EPUB, who is narrating, voice gender, etc.
… with media overlays. synthetic, cloned, human, gender etc.
AvneeshSingh: MTM is using synthetic voice for media overlays.
George: all of these need to consult with.
AvneeshSingh: I will connect with Nordic people.
Morris: noticing with audio books AI voices that are speaking as they are not real. would like to see this in MARC eventually.
… we can identify the voices are artificial.
charles: Schema.org may have already have these values we may need.
<LauraC> I would be careful about indicating voices as “male” or “female.” What is the reason for that? What are people looking for when they make a decision to choose one of those to listen to? What is the quality of voice that is the actual decision point?
Hadrien: 11 Labs has iconic voices. Narrators can provide audio to be cloned. and they get revenue license.
… its not fraud because it could be real and legal.
AvneeshSingh: July 10 next call
<LauraC> I just finished listening to an audiobook that really sounded like AI because the emphasis was odd. But it wasn’t AI because the book was old.