Re: [CODE4LIB] [EXT] Re: [CODE4LIB] A/V and accessibility
I 100% agree. Just for clarity, by requester in my previous email, I meant a person requesting accommodations for the video and not the original persons pushing for the digital collection. The fact is that accessibility remediation is a translation, and different types of remediation can result in information loss just like other translations. Captioning may make the spoken words accessible but may not capture the intonations and other nuances of the dialogue. Transcribing a handwritten letter into electronic text may skip over edit marks and other aspects of handwriting that a researcher may be interested in. Heck, translating handwriting is rarely obvious and can be quite debated. This is why I view special collections and what libraries call archives to be in a different vein than other aspects of accessibility remediation. Making a journal article PDF accessible is mostly about proper markup and reading order (although exceptions and complexities do exist). The main goal is for anyone to be able to read it. But for someone diving into a special collection or archive, their inquiry is different. I've seen historians go on and on about edit marks in letters and marginal notes in books. Each scholar in such works have nuanced inquiries with elements they wish to focus on. To me, making the content accessible to them is about also understanding what they want to access. Most of the time, we think of accessibility as addressing the intersection (dis)ability issues with the content format. However, sometimes we need to add in the further complexity of an individual's actual goals. Personalized accommodations are likely needed. This is the argument I give for our special collections/archives group. Do what is feasible now with current technology and then have a means for providing one-on-one accommodation services. Katherine Deibel | PhD Inclusion & Accessibility Librarian Syracuse University Libraries T 315.443.7178 kndei...@syr.edu 222 Waverly Ave., Syracuse, NY 13244 Syracuse University -Original Message- From: Code for Libraries On Behalf Of Tim McGeary Sent: Wednesday, February 13, 2019 3:45 PM To: CODE4LIB@LISTS.CLIR.ORG Subject: Re: [CODE4LIB] [EXT] Re: [CODE4LIB] A/V and accessibility This is why defining the policy of access is critical. If these digitized collections are intended to be published for the entire public, the needs of the (original) requester is not sufficient; the federal mandates require full accessibility as best to your ability without undue burden. If you aren’t making these available for the entire public, and your policies are well documented about that restriction and the request process, then you have more flexibility to balance the burden of making a collection accessible based on the needs of the specific user. Tim Tim McGeary Associate University Librarian for Digital Strategies and Technology Duke University On Wed, Feb 13, 2019 at 3:37 PM Kate Deibel wrote: > While this is true in the general case, we're again talking about > Special Collections and the needs of the requester. Audio descriptions > are extremely difficult to do as the ideal is to never interrupt other > relevant sounds in the media, especially dialogue. That's a unique > challenge of being precise and fast. My recommendation would be to > make audio descriptions available upon request just as with more quality > captioning. > There is currently no means of automating audio descriptions even of > low quality. AI tools just aren't there yet, and frankly, I'm a little > scared of the idea of a world where AI can view a random scene and > describe what is happening. > > Katherine Deibel | PhD > Inclusion & Accessibility Librarian > Syracuse University Libraries > T 315.443.7178 > kndei...@syr.edu > 222 Waverly Ave., Syracuse, NY 13244 > Syracuse University >
Re: [CODE4LIB] [EXT] Re: [CODE4LIB] A/V and accessibility
This is why defining the policy of access is critical. If these digitized collections are intended to be published for the entire public, the needs of the (original) requester is not sufficient; the federal mandates require full accessibility as best to your ability without undue burden. If you aren’t making these available for the entire public, and your policies are well documented about that restriction and the request process, then you have more flexibility to balance the burden of making a collection accessible based on the needs of the specific user. Tim Tim McGeary Associate University Librarian for Digital Strategies and Technology Duke University On Wed, Feb 13, 2019 at 3:37 PM Kate Deibel wrote: > While this is true in the general case, we're again talking about Special > Collections and the needs of the requester. Audio descriptions are > extremely difficult to do as the ideal is to never interrupt other relevant > sounds in the media, especially dialogue. That's a unique challenge of > being precise and fast. My recommendation would be to make audio > descriptions available upon request just as with more quality captioning. > There is currently no means of automating audio descriptions even of low > quality. AI tools just aren't there yet, and frankly, I'm a little scared > of the idea of a world where AI can view a random scene and describe what > is happening. > > Katherine Deibel | PhD > Inclusion & Accessibility Librarian > Syracuse University Libraries > T 315.443.7178 > kndei...@syr.edu > 222 Waverly Ave., Syracuse, NY 13244 > Syracuse University >
Re: [CODE4LIB] [EXT] Re: [CODE4LIB] A/V and accessibility
While this is true in the general case, we're again talking about Special Collections and the needs of the requester. Audio descriptions are extremely difficult to do as the ideal is to never interrupt other relevant sounds in the media, especially dialogue. That's a unique challenge of being precise and fast. My recommendation would be to make audio descriptions available upon request just as with more quality captioning. There is currently no means of automating audio descriptions even of low quality. AI tools just aren't there yet, and frankly, I'm a little scared of the idea of a world where AI can view a random scene and describe what is happening. Katherine Deibel | PhD Inclusion & Accessibility Librarian Syracuse University Libraries T 315.443.7178 kndei...@syr.edu 222 Waverly Ave., Syracuse, NY 13244 Syracuse University
Re: [CODE4LIB] [EXT] Re: [CODE4LIB] A/V and accessibility
Yeah, it's the domain specific terms that really make or break these systems, especially in academic settings. These might suffice for business domains, but I've seen transcription quality drop quite fast for a STEM class or any non-Western humanities course. Ideally, there would be a feedback loop to these systems but I have yet to see one where you can send in corrections. Katherine Deibel | PhD Inclusion & Accessibility Librarian Syracuse University Libraries T 315.443.7178 kndei...@syr.edu 222 Waverly Ave., Syracuse, NY 13244 Syracuse University -Original Message- From: Code for Libraries On Behalf Of Carol Kassel Sent: Tuesday, February 12, 2019 4:42 PM To: CODE4LIB@LISTS.CLIR.ORG Subject: Re: [CODE4LIB] [EXT] Re: [CODE4LIB] A/V and accessibility Hi everyone, Thank you so much for your replies! I'll reply to each of you individually as well. In answer to your question about which auto-captioning solutions we're looking at, there are 2 main solutions we have our eye on. One is VerbIt and the other is Konch. Both appear to offer reasonable accuracy in the languages we need, though we are still evaluating. Still, as with any of these solutions, they miss some domain-specific vocabulary as well as anything that's mumbled or otherwise hard to understand. Also, we need to figure out our workflow for generating captions/transcripts, getting them into our infrastructure, and allowing for hand corrections and the workflow for revisions resulting from same. The devil is in the details! Best wishes, Carol > > > -Original Message- > From: Code for Libraries On Behalf Of > Carol Kassel > Sent: Monday, February 11, 2019 11:31 AM > To: CODE4LIB@LISTS.CLIR.ORG > Subject: [CODE4LIB] A/V and accessibility > > Hi, > > We're working on a roadmap for making A/V content from Special > Collections accessible. For those of you who have been through this > process, you know that one of the big-ticket items is captions and > transcripts. In our exploration of options, we've found a couple of > pretty good auto-captioning solutions. Their accuracy is about as good > as what you'd get from performing OCR on scanned book pages, which > libraries do all the time. One possibility is to perform > auto-captioning on all items and then provide hand-captioning upon > request for the specific items where a patron needs better captions. > > This idea will be better supported if we know what our peer > institutions are doing... so what are you doing? Thanks to those to > whom I've reached out personally; your information has helped > tremendously. Now I'd like to find out from others how they've handled this > issue. > > Thank you, > > Carol > > -- > Carol Kassel > Senior Manager, Digital Library Infrastructure NYU Digital Library > Technology Services c...@nyu.edu > (212) 992-9246 > dlib.nyu.edu > > > -- Carol Kassel Senior Manager, Digital Library Infrastructure NYU Digital Library Technology Services c...@nyu.edu (212) 992-9246 dlib.nyu.edu
Re: [CODE4LIB] [EXT] Re: [CODE4LIB] A/V and accessibility
Hi everyone, Thank you so much for your replies! I'll reply to each of you individually as well. In answer to your question about which auto-captioning solutions we're looking at, there are 2 main solutions we have our eye on. One is VerbIt and the other is Konch. Both appear to offer reasonable accuracy in the languages we need, though we are still evaluating. Still, as with any of these solutions, they miss some domain-specific vocabulary as well as anything that's mumbled or otherwise hard to understand. Also, we need to figure out our workflow for generating captions/transcripts, getting them into our infrastructure, and allowing for hand corrections and the workflow for revisions resulting from same. The devil is in the details! Best wishes, Carol > > > -Original Message- > From: Code for Libraries On Behalf Of Carol > Kassel > Sent: Monday, February 11, 2019 11:31 AM > To: CODE4LIB@LISTS.CLIR.ORG > Subject: [CODE4LIB] A/V and accessibility > > Hi, > > We're working on a roadmap for making A/V content from Special > Collections accessible. For those of you who have been through this > process, you know that one of the big-ticket items is captions and > transcripts. In our exploration of options, we've found a couple of pretty > good auto-captioning solutions. Their accuracy is about as good as what > you'd get from performing OCR on scanned book pages, which libraries do all > the time. One possibility is to perform auto-captioning on all items and > then provide hand-captioning upon request for the specific items where a > patron needs better captions. > > This idea will be better supported if we know what our peer > institutions are doing... so what are you doing? Thanks to those to whom > I've reached out personally; your information has helped tremendously. Now > I'd like to find out from others how they've handled this issue. > > Thank you, > > Carol > > -- > Carol Kassel > Senior Manager, Digital Library Infrastructure NYU Digital Library > Technology Services c...@nyu.edu > (212) 992-9246 > dlib.nyu.edu > > > -- Carol Kassel Senior Manager, Digital Library Infrastructure NYU Digital Library Technology Services c...@nyu.edu (212) 992-9246 dlib.nyu.edu
Re: [CODE4LIB] [EXT] Re: [CODE4LIB] A/V and accessibility
UNT Digital Libraries and the Portal to Texas History are starting to test the waters here too with a ton of content to catch up on. Early days. Vendors: We've tested 3Play and then rev.com. At the latest Accessing Higher Ground (AHG) conference, the latter was getting talked up a lot by ODA office folks as their current preferred vendor given speedy turnaround and cost ratio. Automation: I've played with https://github.com/agermanidis/autosub with decent~ish output given a few test cases. I know there are a few amazon-related demos out there too. No formal workflows on my end yet, but I think your outlined approach is generally what my preferred option would look like too. Hope to hear more from you/others on what they are trying. digression: I note a handful of folks I talked to at AHG didn't think OCRing text in image content was good enough for real compliance when they saw the gibberish it often spits out, which would lead me to believe automated efforts for A/V would leave us open to the same sorts of complaints, but we do what we can, right?). Also captions/transcriptions are only going to get us 1/2 way to WCAG AA given the need for audio-descriptions. Maybe text-to-speech here? 3play has a plugin along those lines. Other issues on my plate: looking at cleanup and audio-desc. Script authoring (probably will use WGBH Cadet); other outliers like doing webvtt chapters; what webvtts should look like for music where you want to give substantial info (i.e movements in symphonies, describing affect in a French Aria, or audio-describing a performance with something better than "[jazz music playing]"; and tangentially to your original question: what does it look like to hire/contract and ASL signer to make derivative files to meet that need if/when it comes up. As to storage, our webvtts are going into a local gitlab repo, and then we have a few local scripts to push them onto public DL filesystem. I haphazardly dream of a future scenario where the DL public interface provided links from automated transcripts to the git repo for some sort of crowdsource cleanup effort. Side note: ODA office folks looked at me with a lot of puzzlement when I asked how they were archiving/storing captioned media! For now at least, non captioned A/V have links in their descriptive records to make requests, which we'll typically honor ASAP with a vendor supplied file. https://texashistory.unt.edu/ark:/67531/metadc700196/ (see sidebar for request link). For now this just populates a simple webform with some boilerplate. Interested if you can share more of what you are up to. Cheers, William Hicks Digital Libraries: User Interfaces University of North Texas 1155 Union Circle #305190 Denton, TX 76203-5017 email: william.hi...@unt.edu | phone: 940.891.6703 | web: http://www.library.unt.edu Willis Library, Room 321 On 2/11/19, 4:02 PM, "Code for Libraries on behalf of Goben, Abigail H" wrote: I can't speak to captioning but I use temi.com for my transcription for the class that I teach. It's .10 a minute, it's machine-transcription. Overall it does a really decent job and I can't argue with the price. The transcription takes about half the time of the video, I do light editing, and post. -- Abigail H. Goben, MLS Associate Professor Information Services and Liaison Librarian Library of the Health Sciences University of Illinois at Chicago 1750 W. Polk (MC 763) Chicago, IL 60612 ago...@uic.edu -Original Message- From: Code for Libraries [mailto:CODE4LIB@LISTS.CLIR.ORG] On Behalf Of Kate Deibel Sent: Monday, February 11, 2019 1:37 PM To: CODE4LIB@LISTS.CLIR.ORG Subject: Re: [CODE4LIB] A/V and accessibility I'd love to hear what auto-captioning options you've found to be tolerable? What I can say is that this is the informal policy I've been promoting for accessibility in our special collections. In general, any accommodation requests in special collections will likely be part of a very nuanced, focused research agenda. Thus, any remediation will likely not only have to be specific to the individual's disability but also the nature of their research. In the case of A/V, a rough transcription may be enough if they are focusing more on the visual side of it. For others, though, a more thorough transcription may be required. All in all, your approach sounds wise. Katherine Deibel | PhD Inclusion & Accessibility Librarian Syracuse University Libraries T 315.443.7178 kndei...@syr.edu 222 Waverly Ave., Syracuse, NY 13244 Syracuse University -Original Message- From: Code for Libraries On Behalf Of Carol Kassel Sent: Monday, February 11, 2019 11:31 AM To: CODE4LIB@LISTS.CLIR.ORG Subject: [CODE4LIB] A/V and accessibility Hi, We're working on a roadmap for making