In the context of increasing the accessibility on PrivacyGuides, I was wondering whether the community would see value in adding text to speech option to our recommendations and knowledgebase articles.
The reason for doing this would be to provide users with poor eyesight (or users which are blind) a new way to access the education we provide.
The first question in my head is whether this is worth persuing at all. To my knowledge blind people have screen readers which would help them, but I have no clue in how effective those are in a web browser.
If we do decide to go ahead, we would have to decide how we will implement this.
Currently I see two possible models with each their ups and downs.
The first model would be to hire someone to read and record everything. This comes with the upside of being able to make the recording sound like an actual human. This makes it more enjoyable to listen to.
The downsides would be that we would have to pay someone, which means we cannot use those funds for other projects. Another downside would be that its hard to maintain. With ever single word change on the site at least a part of the recording will have to be redone.
The other model would imply using a text to speech service. This comes with the upside of being way cheaper compared to hiring someone, and being more easy to maintain. Some downsides would be that it can sound rather robotic, and some pages might be hard for a text to speech engine to understand, like our dns recommendation table.
A last question would be: if we go ahead, do we want to implement this on the translated pages as well in those native languages, or do we stick to the main english version.
The audio seems not so high quality and slow. I think that most people with vison issues are used to hearing things a lot faster. Not sure how useful this is but that question I would like to add. What should be the format of the audio?
Just realized there probably must be standards for this.
I like this idea from an accessibility perspective, but I don’t think this is the way to do it personally.
For just TTS I think there are better client-side programs people will be able to use to read the website than what we’ll be able to provide. We should do more testing with a11y tools for sure, but I think that our site should be basic enough HTML for any screen reading tools to handle properly.
I think that if we wanted to create A/V content a better use of our time would be to create videos similar to, for example, https://youtube.com/@techquickie with dedicated scripts that are more optimized for listening than the website is. These could be as simple as just talking to a camera, or just voiceover-only videos with graphics. Either way it’s probably just a few hours per video and it’d reach a whole new demographic as well.
Video would still leave the question about maintainability right? Can you alter specific sections of a youtube video after its published? We would also have to decide on a platform, like do we want to push people to a google platform like Youtube?
The techqhuickie style videos sound like a great idea. Although I think maybe adding a local TTS recommendation category for people who are helping family and friends set up their devices may be a good option. I spent around an hour yesterday looking for a good TTS service to use on GrapheneOS and ended up going with google TTS because everything else I could find seemed like it hadn’t been updated in years.
YouTube & PeerTube probably. Can’t really edit a video, although you could remove sections, but I’m thinking each video would probably just cover a single topic and be short enough that we could just unlist it entirely if things change.
I can see that working for our knowledgebase articles, but how about the recommendations, does do change quite often, sometimes even for quite big reasons, like with Raivo for example.