Free blind browser?

[Frogleg]

Is there one? I've been googling and judging by some prices mentioned,
it seems unlikely.

Discussion of 'alt' text and a little reading got me interested in
"seeing" how my sites sounded.

 

[Andrew Urquhart]

*Frogleg* wrote in alt.html:

[Re: Free blind browser?]
Is there one? I've been googling and judging by some prices mentioned,
it seems unlikely.

Discussion of 'alt' text and a little reading got me interested in
"seeing" how my sites sounded.

I'm guessing you mean for Windows?

Not free (but will run for 40 minutes at a time) and is not a browser as
such but is optimised for IE:
http://www.freedomscientific.com/fs_downloads/jaws_form.asp

In addition (and not limited to Windows)
Text browsers such as Lynx and Opera's various text modes should also be
of use. I'm also reading rumours about the up-coming audio capabilities
of Opera 7.6 so you might want to check that out too.

 

[Karl Groves]

Is there one? I've been googling and judging by some prices mentioned,
it seems unlikely.

Discussion of 'alt' text and a little reading got me interested in
"seeing" how my sites sounded.

Take a look at http://usabilityinfo.com/tools.php?viewonly=dtopapp

-Karl

 

[Chris Morris]

Is there one? I've been googling and judging by some prices mentioned,
it seems unlikely.

Discussion of 'alt' text and a little reading got me interested in
"seeing" how my sites sounded.

I don't know of a (good) free one, but you can download trial versions
of them - I'd recommend the 30-day IBM Home Page Reader trial - use it
a lot in that time and you should get a fair idea of how speech
browsers work. After you've got the idea you can probably use a text
browser as a substitute - there *are* differences, and important ones,
but if you're aware through usage of them, you can deal with that.

 

[Toby Inkster]

I'm also reading rumours about the up-coming audio capabilities
of Opera 7.6 so you might want to check that out too.

It's not just rumours...
http://www.opera.com/pressreleases/en/2004/07/26/

(Though I do hope they extend support beyond just Win2K/WinXP. It probably
depends on what IBM do though -- IBM discontinued ViaVoice for Linux a
couple of years ago, but IBM are now pushing Linux on the desktop
internally, so who knows what may happen?!)

 

[Frogleg]

Thanks to all. I will try your suggestions.

 

[Frogleg]

I don't know of a (good) free one, but you can download trial versions
of them - I'd recommend the 30-day IBM Home Page Reader trial - use it
a lot in that time and you should get a fair idea of how speech
browsers work.

41Mb...over dialup...whew! But worth it. Learned one thing right off.
Not to use abbreviations in addresses. "123 Maple Avenue" doesn't take
much more space/time than "123 Maple Ah-vay (Ave.)." It would be nice
if one could teach it (HPR) that Wythe is pronounced "with," but then
few people get it right when reading it, either.

Fun, as it makes my sites sound important -- like being read on the
radio as worthwhile news.

I found some interesting material here:
http://www.webaim.org/techniques/images/5

I think I may have some trouble trying to decide exactly what the text
should be. How much does an unsighted person want to know about a
photograph? I tend to over-explain about *everything*, but which is
appropriate? "Photo of creek" or "Large landscape photo of Jones
Creek at high tide"? I will try to avoid "Large landscape photo of
Jones Creek with an egret over there to the left."

 

[Chris Morris]

I think I may have some trouble trying to decide exactly what the text
should be. How much does an unsighted person want to know about a
photograph? I tend to over-explain about *everything*, but which is
appropriate? "Photo of creek" or "Large landscape photo of Jones
Creek at high tide"? I will try to avoid "Large landscape photo of
Jones Creek with an egret over there to the left."

Appropriate alt text for photographs is always a tricky one - it
depends why the photograph is there.

For a photograph album, the second would probably be the best
available (though perhaps a variant of the third if there's lots of
photos describable by the second). For a decorative image on a web
page, possibly alt="". For a page about the wildlife of Jones Creek,
perhaps "At high tide, the egrets congregate on the left of the
creek".

It really depends on the context that the image is being used in -
think what text you'd put there if you couldn't use images [1], and
put that text there.

[1] For a photograph album, perhaps it's best to think
<img src="a.jpg" alt="a"> is equivalent to <a href="a.jpg">a</a>.

 

[Neal]

It would be nice
if one could teach it (HPR) that Wythe is pronounced "with," but then
few people get it right when reading it, either.

Is it Welsh? What happens if you set it as a span with a language
attribute?

 

[jake]

In message <[email protected]>, Frogleg

41Mb...over dialup...whew! But worth it. Learned one thing right off.
Not to use abbreviations in addresses. "123 Maple Avenue" doesn't take
much more space/time than "123 Maple Ah-vay (Ave.)." It would be nice
if one could teach it (HPR) that Wythe is pronounced "with," but then
few people get it right when reading it, either.

You can do. Quite simply.

You need to go to Settings --> Speech --> Dictionary --> Brit/US English

and then put an entry in the 'Special Words' list.

You'll need to read the documentation to find out about the phonetic
spellings and stress patterns.

Example 'Site Map' is often spelled 'Sitemap' on many Web pages, and HPR
pronounces this as something like SIH-TEE-MAP.

I have the following entry in my dictionary that makes it pronounce it
'SITE MAP', with the stress on the 1st syllable.

`[.1sYt] `[.2mAp]

That's a fairly simple substitution. You also have control over (I seem
to remember) pitch, timing, stress, and other things if you really need
to.


[snip]

regards.

 

[jake]

Is it Welsh? What happens if you set it as a span with a language
attribute?

Depends.

If HPR is set to 'Automatic Language Detection', it will attempt to
pronounce the word according to whatever language is specified in the
<span></span>.

If HPR is set to a specific language (the normal setting) it will ignore
any 'lang' attribute and pronounce it according to the language it's set
to.

(BTW Welsh isn't an HPR supported language).

Of course, if there's an entry in the 'special words' dictionary, it
will take that preference into account.

regards.

 

[Jukka K. Korpela]

If HPR is set to a specific language (the normal setting) it will
ignore any 'lang' attribute and pronounce it according to the
language it's set to.

Sounds bad. I don't think it did that when I tested it, but then again,
I often remember things wrong. Of course a speech browser should have
'lang' support enabled by default, unless it's very experimental - and
IBM Home Page Reader has had very good support to several languages long
ago.

 

[Frogleg]

Is it Welsh? What happens if you set it as a span with a language
attribute?

Might be, originally. Around here (SE Virginia), it refers to George
Wythe, a figure in Colonial gov't and the Revolutionary War.

 

[Frogleg]

In message <[email protected]>, Frogleg

You can do. Quite simply.

You need to go to Settings --> Speech --> Dictionary --> Brit/US English

and then put an entry in the 'Special Words' list.

<snip>

Cool. Can I just enter "Wythe == with"? (joke!)

Just going over a couple of sites with HPR has made me aware of things
that reading about didn't make nearly as clear.

 

[jake]

Sounds bad. I don't think it did that when I tested it, but then again,
I often remember things wrong. Of course a speech browser should have
'lang' support enabled by default, unless it's very experimental - and
IBM Home Page Reader has had very good support to several languages long
ago.

I've just re-tested by getting it to speak:

<span lang="es">Quiero ir al trabajo</span>
<span lang="fr">Je ne mange pas</span>
<span lang="de">Ja wohl</span>

If HPR is set to 'British English', then it makes an attempt to
pronounce them as if they were 'English' words; if I set it to
'automatic language detection' it pronounces them all with the correct
pronunciation (and accent).

Still, one thing that is quite cleaver is that (when in 'automatic
language detection' mode) if it reads a page where the author has failed
to include a 'lang=' on, say, <html> then within half-a-dozen words it
locks onto the correct language and begins speaking with the correct
pronunciation and accent.

regards.

 

[Jukka K. Korpela]

Still, one thing that is quite cleaver is that (when in 'automatic
language detection' mode) if it reads a page where the author has
failed to include a 'lang=' on, say, <html> then within half-a-dozen
words it locks onto the correct language and begins speaking with the
correct pronunciation and accent.

Interesting! So they have mixed up two things, heuristic language
recognition and utilization of declared language information - both fall
under "automatic language detection". (Both of them are potentially very
useful, but they are different things.)

I wonder why HPR doesn't do any lookahead to recognize the language
_before_ it starts speaking. After all, mispronouncing the start of a
document tends to obscure the most important part of the content.

 

[jake]

<snip>

Cool. Can I just enter "Wythe == with"? (joke!)

[snip]

Yes, you can (just tested it).

regards.

 

[jake]

Interesting! So they have mixed up two things, heuristic language
recognition and utilization of declared language information - both fall
under "automatic language detection". (Both of them are potentially very
useful, but they are different things.)

I wonder why HPR doesn't do any lookahead to recognize the language
_before_ it starts speaking. After all, mispronouncing the start of a
document tends to obscure the most important part of the content.

I'm not sure why. It seems to use the contents of <title></title> to
recognise the language, using an 'English' pronunciation to speak it in
the meanwhile.

Still, I guess that from a practical point of view it's not really an
issue.

If my language is French, then presumably I would want to read French
pages -- and therefore have the browser set to read French as default.

If I was multi-lingual and wanted to browse pages in 2 or 3 languages
then this, I assume, is where 'automatic language detection' would come
in handy.

The only thing that does seem wrong is that if the default is anything
other than 'automatic language detection', then 'lang' is ignored.

regards.

 

[Isofarro]

It's not just rumours...
http://www.opera.com/pressreleases/en/2004/07/26/

(Though I do hope they extend support beyond just Win2K/WinXP. It probably
depends on what IBM do though -- IBM discontinued ViaVoice for Linux a
couple of years ago, but IBM are now pushing Linux on the desktop
internally, so who knows what may happen?!)

Back in July 2004 IBM licensed its text to speech technologies, particulary
for use on Linux <url:http://www.tmcnet.com/usubmit/2004/Jul/1055923.htm>

Accessibility on Linux tends to be heading towards Sun's ATK.

 

[Deryck]

41Mb...over dialup...whew! But worth it.
Yes.

Learned one thing right off.
Not to use abbreviations in addresses. "123 Maple Avenue" doesn't take
much more space/time than "123 Maple Ah-vay (Ave.)." It would be nice
if one could teach it (HPR) that Wythe is pronounced "with," but then
few people get it right when reading it, either.

I have learnt a lot from this in just a couple of hours.
Do you know if it is possible to have it read large numbers (such as phone
numbers or ISBN numbers) as a sequence of digits rather than as "millions
and thousands"?

Cheers

Deryck