This test is a follow-up to my original slider test, in response to the question "What happens if the values of aria-valuenow and aria-valuetext are extremely different (not only in how they're expressed, but the numerical value)?" In this example aria-valuemax="50" and aria-valuenow is a value between 1 and 50, whereas aria-valuetext is an expression of this value as a percentage. I also disabled sliderValue updates (the span that follows the slider) to be sure screen readers aren't reading that text.
Here's a link for testing tab order.
Sorry, mousers. This slider wasn't designed for you.
Dragon NaturallySpeaking 13 (Windows) now supports aria-label, so "Click" followed by the value of aria-label works to give the slider focus. However, this isn't true for older versions of Dragon, but they do support the title attribute for this same purpose, and it's better for speech recognition users anyway since it's visually discoverable. But do screen readers all support title if no aria-label is present?
If both title and aria-label are present, JAWS announces both (in both browsers), which is redundant.
Conclusion: It's reasonably safe to use title rather than aria-label.
If both of these attributes are present, what value does each screen reader announce when the slider has focus, and what value does it announce as the user change the slider position?
Given the problem with NVDA in IE11, what does NVDA read if aria-valuenow is removed, and how does that effect other screen readers?