WebAIM - Web Accessibility In Mind

E-mail List Archives

Re: screen reader versions for testing

for

From: Birkir R. Gunnarsson
Date: Oct 30, 2016 2:40PM

The reason I have not encouraged testing with ChromeVox is that it is
very rarely used.
According to the latest WebAiM screen reader user survey:
http://webaim.org/projects/screenreadersurvey6/
it is the primary screen reader for 0.3% of correspondants, while NVDA
is in the 15% range (and much higher if you take secondary screen
readers into acount).
Sadly it is not enough to make sure the webpage code conforms to
standards, it needs to be tested with at least one assistive
technology, usually a screen reader, and that usually means learning
and implementing some workarounds to address the qwerks of that
particular screen reader.
If I am putting an effort into that, I want to make sure to use a
popular screen reader, so those workarounds are noticed.
Of course screen reader usage pattern changes, and we all should keep
a close eye on the WebAIM survey (and other usage statistics if they
become available).
A thumbs up for WebAIM for taking the initiative to carry out this
survey. It is incredibly valuable when recommending and formulating a
corporate accessibility testing strategy, management wants
justification and numbers behind all recommendations.
The Android/Talkback development is exciting and I am keeping a close
prosthetic eye on it, in case it surpasses Voiceover use on responsive
web in the near future, it could maybe do that, seeing as Google is
doing good while the latest Apple upgrades are a bit underwhelming
(well, in my personal opinion that is).
-B


On 10/30/16, Kevin Chao < <EMAIL REMOVED> > wrote:
> I've been doing lots of a11y testing using ChromeVox Next
> <http://www.chromevox.com/next.html>; and TalkBack
> <https://play.google.com/store/apps/details?id=com.google.android.marvin.talkback&hl=en>
> with
> Chrome. I've found it to be comparable to/better than Mac/iOS VoiceOver. In
> the past half year, there have been lots of excellent improvements to
> Google's screen readers and browsers, so strongly recommend for these to be
> factored in AT test matrix.
>
> On Sun, Oct 30, 2016 at 4:20 AM Birkir R. Gunnarsson <
> <EMAIL REMOVED> > wrote:
>
>> We generally test with NVDA (current - 1) with Firefox (current -2).
>> NVDA is free, open source (so available to the general user at no
>> cost), has good visual tools to help developers and does not hide
>> accessibility issues like Jaws does (I appreciate Jaws trying to fill
>> in the gap for the end users but it makes it a bad tool for testing).
>> Since I am heavily involved in development and testing of contet, I
>> sanity check it with Jaws and IE, and we try to file bug and work
>> around the most critical problems we see occurring in that
>> combination.
>> For responsive web, we use iOS, latest (because upgrading is easy),
>> iPhone 6 in portrait mode (testing in portrait and landscape on phone
>> and tablet adds a lot of overhead very quickly).
>> Generally, banks recommend that users upgrade to latest versions of
>> browsers for security reasons.
>> We are looking into testing at least key pages with screen
>> magnification and speech recognition as well.
>>
>> Of course we focus primarily to make sure our code validates and that
>> our ARIA, when we use it, is correct.
>> Cheers
>>
>>
>>
>>
>> On 10/29/16, JP Jamous < <EMAIL REMOVED> > wrote:
>> > Here we test with the latest versions of JAWS/Internet Explorer,
>> > NVDA/Firefox and Voiceover/Safari.
>> >
>> > It makes it a bit hard to find the happy medium as all 3 screen readers
>> > render HTML markup differently. To achieve the happy medium, we try to
>> focus
>> > on proper semantic whenever we can. Sometimes that is not possible and
>> > we
>> > notice that NVDA and Voiceover tend to behave similarly, but JAWS is
>> > different since it drills deeper into the markup.
>> >
>> > We do test every now and then with older versions of the 3 screen
>> readers in
>> > case we run into an issue. As a good example, aria-describedby and
>> > aria-labelledby were not supported with Voiceover on iOS 10. We tested
>> our
>> > code against iOS 9.4 and found that it worked fine. That was when we
>> > realized that it was a bug on behalf of Apple.
>> >
>> >