rns.recipes

hello everyone!

I am really excited about Reticulum and trying to help some of my friends and comrades get involved. many of them are intimidated by the perceived technical complexity of Reticulum, so I am trying to write a simple user's guide for them that will explain building their own networks step-by-step. some of these friends and comrades rely on screen-readers or speech-to-text to navigate their devices, so I am trying to learn how to use a terminal and various Reticulum apps using those programs, so that I can pass along clear and simple instructions. in particular I intend to develop instructions for using Reticulum with the Orca screen-reader, the NVDA screen-reader, and Talon speech-to-text. so far, I have only tried using Orca myself, on Ubuntu.

I was able to use the terminal with Orca pretty easily, but I had trouble with the Reticulum apps that I tried. NomadNet was super easy to navigate with just my keyboard, which was awesome, but I couldn't figure out how to read aloud the shortcuts at the bottom of the screen (ex. [C-w] Disconnect). and I could not figure out how to get Orca to read aloud from the Micron pages that I was able to display.

Orca did not work with Sideband at all. It did not even say anything when I opened or closed the application.

MeshChatX was more challenging to navigate with my keyboard, although it was possible. Orca did read aloud all of the text and options that I could see in MeshChatX. but I still had trouble with the Micron pages; it wouldn't read the body of them, but I could tab from link to link and Orca would read the link text aloud. I have no idea why.

so, here are my questions:

is there anyone on this forum already using Reticulum with a screen-reader, speech-to-text, or any other adaptive tech? if so, do you have any advice or insight to share?

does anyone know of any documentation about accessibility for Reticulum or any of the main Reticulum apps? I checked the manual, the community wiki, these forums, and the docs for NomadNet, Sideband, and MeshChatX, and I couldn't find anything. if there is any existing guidance for app compatibility with screen-readers or other adaptive tech, that would be so helpful to read.

does anyone have any other guidance on what I should do next? I am uncertain whether to attribute my difficulties to my own knowledge gaps, Reticulum bugs, Orca bugs, something else. any clarity that anyone can provide on the subject would be very welcome.

thank you all for your time!

Very interesting, whole classes of device could be designed for this use case. A voice paging system over lxmf comes to mind. No visual user interface, just physical buttons. Both could be used, text to speech and voice messages. I love retibooks but i dont think ill ever use an e reader on the device I want to use for RNS, but TTS would make that possible.

Are there screen readers that use OCR?

(Still new to Reticulum and don't use a screen reader) I don't think TUIs have a good track record for screen-reader-friendliness. I've heard of blind people liking Dungeon Crawl Stone Soup which uses ncurses(?) but idk what they use to access it. Is there a (local) web frontend people here recommend? Not that those can't also be an accessibility nightmare, but probably less so? If not, I wonder how far you can get with plain old cli commands. Is there a particular usage you're focused on (direct messaging, chatrooms, pages, etc)?

Alledegly_noderunner wrote:

Very interesting, whole classes of device could be designed for this use case. A voice paging system over lxmf comes to mind. No visual user interface, just physical buttons. Both could be used, text to speech and voice messages.

I do love the idea of a pager! And I am very curious what kinds of possibilities exist for dual use of TTS and STT, although I haven't necessarily seen them used together myself. I doubt we will see good integration between these two types of systems until there are really good open source STT navigation programs, though.

Zenith wrote:

Are there screen readers that use OCR?

Not to my knowledge. From everything I've learned so far, OCR programs handle the translation of an image of text into text, while screen-readers translate text into audible speech. I think that part of the reason why they are not generally integrated is that OCR translations may not always be accurate, so it's best practice for those translations to be reviewed before they are published for use with screen-readers.

Base64Keymashing wrote:

I don't think TUIs have a good track record for screen-reader-friendliness. I've heard of blind people liking Dungeon Crawl Stone Soup which uses ncurses(?) but idk what they use to access it. Is there a (local) web frontend people here recommend? Not that those can't also be an accessibility nightmare, but probably less so? If not, I wonder how far you can get with plain old cli commands. Is there a particular usage you're focused on (direct messaging, chatrooms, pages, etc)?

That is actually really surprising to me! I had assumed that text-based user interfaces would be easier to make screen-reader friendly by virtue of already being all-text, with no images or icons to translate for a screen-reader. I'll definitely be doing some more research.

I am focused on finding the simplest possible means by which a user with a screen-reader could use the full range of functionality that Reticulum applications offer to users who don't rely on screen-readers, and on identifying exactly what accessibility needs each Reticulum app does or does not support. If no Reticulum apps are compatible with any screen-readers, then I will make a guide that explains how to use exclusively the CLI; if some Reticulum apps are compatible with some screen-readers, then I will make guides for both; if there is any Reticulum app that can be made compatible with the major screen-readers, even with some configuration, then I will make a guide for that app. That goes for each feature: direct messaging, pages, chatrooms, etc.

It looks like urwid (TUI framework I see in the nomadnet import list) has the ability to do some sort of html output. (Search for Other Display Modules in their manual) Maybe that would be a good starting point. I can try to mess around with it tomorrow.

Base64Keymashing wrote:

It looks like urwid (TUI framework I see in the nomadnet import list) has the ability to do some sort of html output. (Search for Other Display Modules in their manual) Maybe that would be a good starting point. I can try to mess around with it tomorrow.

thanks so much for the suggestion! is this the manual section you were referring to? I've read it over a couple times now, but if I'm honest a lot of it is going over my head. if there is a way to use html output though that sounds like it could be very helpful! I don't know whether Orca is compatible with every browser but I do know it is intended to be able to read html.

I have experimented a bit more myself with Orca and NomadNet, answering some of my own questions and generating new ones in the process. for starters, I have determined that the command list on the bottom of the screen can be read aloud with the right and left arrow keys. what still seems unpredictable to me is where your focus needs to be before you hit those arrows in order to get each list of commands. most pages have more than one list, and which one is visible depends on what you have in focus / what you're trying to do, but when using the left and right arrows it is not always clear to me exactly where the focus is or where it's being moved to.

I believe that my confusion can be at least partially attributed to the fact that it seems Orca cannot necessarily distinguish between the left-side and right-side panels. where I was tabbing through things like my list of saved nodes relatively quickly on my first try, I am now going much shlower and waiting to see whether anything is read aloud after the name of the saved node has been fully read. doing this, I learned that Orca will read straight across the screen, from the beginning of the name of a saved node to the end of a line of a Micron page level with that line of text. it does the same thing in the Guide section, too, so that the Introduction reads aloud like this:

Introduction Concepts & Terminology Communicate Freely. Channels & RRC Interfaces The intention with this program is to provide a tool that allows you to build Hosting a Node private and resilient communications platforms...

so, it seems that Orca can read Micron as it is displayed in NomadNet after all. the trouble is distinguishing between the different sections of the screen displaying information, and navigating between them without consistently clear audio feedback. I'm kinda hoping that if I just keep messing around and get really familiar with the key navigation, I'll be able to learn the pattern underlying the navigation logic enough to explain it to someone else. I'm planning to experiment more in MeshChatX with Orca when I get a chance, with the same hope.

Yep, that's the link! Sorry for not pasting it, MeshchatX on Android doesn't let me paste in the browser for some reason. It's a pretty short entry, I'm not 100% sure what it does either, but I'm hoping a more structured output than the sidebar reading you've encountered. Another thing as you're working on this: have you asked the potential users of these features in your community what they prefer for tech? Like, screen reader or OS? I'd be testing with Orca too since that's what I have, but I know people using a screen reader can have strong opinions about OS/screen reader and different screen readers have different quirks. It's hard to test different ones on your own ofc and testing with Orca's a good start, but just be sure not to commit too hard to it without checking.

Just chiming in to say that accessibility is very important. Sideband uses Kivy, which seems to totally lack accessibility features unfortunately. One other issue is that NomadNet selects the first letter of buttons instead of the whole text, so Orca only reads a single letter.

Zenith wrote:

Are there screen readers that use OCR?

I'm quoting you again because I just discovered that my first answer to your question was incorrect. Some screen readers do have an OCR function, and NVDR is one of them. From what little I have read so far, it is not always reliable with longer or more complicated documents, and it is still ideal for content to not require OCR to be readable, but it's there. I haven't found anything suggesting that Orca has an OCR function, but I also still haven't finished reading the Orca documentation yet, so I could be wrong on that too.

Base64Keymashing wrote:

Yep, that's the link! Sorry for not pasting it, MeshChatX on Android doesn't let me paste in the browser for some reason. It's a pretty short entry, I'm not 100% sure what it does either, but I'm hoping a more structured output than the sidebar reading you've encountered. Another thing as you're working on this: have you asked the potential users of these features in your community what they prefer for tech? Like, screen reader or OS? I'd be testing with Orca too since that's what I have, but I know people using a screen reader can have strong opinions about OS/screen reader and different screen readers have different quirks. It's hard to test different ones on your own ofc and testing with Orca's a good start, but just be sure not to commit too hard to it without checking.

No worries about the link! And yes, that's how NVDR and Talon landed on my list. I do not personally know anyone that uses Orca, but since I've never used a screen-reader myself before, I wanted to practice and get familiar with one on the OS most familiar to me first. I plan to use a virtual machine to test NVDA and Talon on Windows after I get used to using Orca and Talon on Ubuntu.

Anonymous wrote:

Just chiming in to say that accessibility is very important. Sideband uses Kivy, which seems to totally lack accessibility features unfortunately. One other issue is that NomadNet selects the first letter of buttons instead of the whole text, so Orca only reads a single letter.

I definitely agree that accessibility is very important! That's unfortunate to hear about Sideband. And I noticed what you're talking about on the NomadNet navigation bar, but I must have missed it on the other buttons. I was wondering if it might have been intentional, because I have heard some users prefer first-letter navigation, but that seems like the kind of feature that would be most useful as a custom configuration rather than a default.

I'll share another update with everyone whenever I get the chance to mess around with this a bit more.

I am focusing on improving accessibility in MeshChatX for future versions, thanks for bringing this up.

micron-parser-js now has initial ARIA tag support for screen reader accessibility in version 1.0.5, in addition to numerous bug fixes.

The demo has the accessibility option enabled by default which you can try with any screen reader or extension here:
https://rfnexus.github.io/micron-parser-js/

Also there's now a Serif font with Noto Serif Nerd Font provided as the default. Anything in a literal is still rendered in monospace, but text content can be rendered in serif. I've also improved support for the browsers font size and zoom.