Comment 1 by jteh on 2015-04-09 07:29
The current architecture doesn't allow for this, as speech only gets the text. However, stuff like this might (and that's a big might) be possible after #4877.

Comment 2 by jmuheim on 2015-04-09 07:31
An idea is just popping into my mind: why not having an HTML output of the speech viewer, where each output has special classes related to the type of output?

For links, this could look like this:

this is the link content

Or a heading 1:

this is the heading content

Using CSS3 magic, this would allow to create really great visual representations of the linear page flow:

.heading-1::before {
display: block;
content: 'Heading 1: '
color: blue;
}

Using JavaScript one could even look out for stuff like skipped heading levels and mark them visually, etc. I'm sure this would encourage a lot of people to create and share their own CSS/JS enhancements to the output which themselves would boost accessibility testing efficiency.

Comment 3 by jmuheim on 2015-04-09 07:47
I have just checked out the Unspoken plugin. While buggy, this looks quite interesting, and it seems to be possible to direct the speech output into another direction somehow.

Comment 4 by Palacee_hun on 2015-04-09 10:17
Please note that mainly two (yes two) developers work on entire NVDA code, tens of thousands of lines of code, sometimes very complex. While of course I understand and value your request of colorized or HTMLized output in speech viewer, I am afraid that this would be too big of a work for the two developers and thus wouldn't fit in their schedule. Even commercial, sometimes very expensive screen readers with considerably larger developer teams don't have such a feature (as far as I know).
The addon aproach would be far better in my view, or even a special build of NVDA for web accessibility testing with specific features for that. But in either case extra developers must be involved. If you can and willing to develop some code related to this (let them be some patches or even code templates), then please do and share.

Cc @feerrenrut: being a sighted NVDA developer, how do you value this request?

Currently, when things like "link" are announced they are output to speech viewer as well. I would not want to remove this mode of operation, since it's useful for understanding or recording the true UX for someone relying on speech alone. Assuming we are talking about replacing text like "link" with a style, I would suggest the addition of a mode where this metadata is rendered in different styles. However, from an accessibility testing point of view, I am concerned that mode would allow testers to actually ignore the core user experience.

I think I need to better understand the use case here.

I still think that this is a very good idea.

In general, for users new to accessibility testing, it can be hard to distinguish between "meta info" and "real content". For debugging purposes and in general an easier way for developers to hop into accessibility testing, this would be very useful.

Let me know if I can be of any additional help.

it can be hard to distinguish between "meta info" and "real content".

@jmuheim That's basically my point, the "meta info" is just as important as the "real content". Wouldn't this make it all to easy to ignore problems caused by the semantics of the information being lost?

I suppose this is really a request for a tool focused on accessibility testing, rather than another way to process the output of NVDA. Excuse me if I'm telling you things you already know, but for those unaware and reading this issue: the output of the speech viewer changes (as does the actual speech), depending on what options you have selected in NVDA. For instance, you may turn off the reporting of links, or lists, etc.