[Blindmath] "Spoken Mathematics on the Web" - Chromevox

[Michael Whapples]

Thanks for this summary Susan, I possibly was going to look at it but in 
general I am not very taken by Google stuff (so very little interest in 
adopting Google chrome for my browser) and I think I had heard someone 
already say that while chromevox does read some MathML they noticed a 
few gaps. Also I am far from taken by the idea of needing to use a 
different assistive technology when I want to browse the web, my main 
screen reader is what I use to access the computer, and there is 
something said for keeping it consistant for all applications (I would 
prefer to concentrate on what I am doing rather than concentrate on how 
to access the information).

Just a few observations and questions of what you say.

Using prosody to assist with understanding equations: I believe it can 
be useful, but I agree that to rely on the prosody alone would be a 
mistake. I think well placed pauses can help with identifying blocks of 
the equation. Also it appears that Apple in their IOS7 stuff do use 
prosody, VoiceOver raises the pitch when speaking a superscript and 
lowers the pitch for subscript, although it does still speak where these 
begin and end. Possibly more useful than prosody though is the ability 
to explore an equation, some can just be too long to process in one go.

The chromevox speech API: I possibly should check this, but I have a 
suspicion that this too will not turn out to be new. Some of this stuff 
I hear about chromevox sounds very familiar when thinking back to the 
firevox extension (http://firevox.clcworld.net) for the firefox browser, 
which also did support some MathML speaking and a speech API (I think it 
was termed as speech CSS or something like that). As the author of the 
firevox extension I believe is involved in Google accessibility (I have 
seen him pop up in a few Google accessibility related videos) I would 
not be surprised if much of the implementation was based upon that. As I 
think the speech API support in firevox goes back quite a few years 
(well over 6, not sure if as far back as 10), one might say what will 
change things now which will make people use this API in webpages?

Nemeth MathSpeak: Thanks for that information, it certainly makes things 
clearer for me how MathSpeak fits in with everything else. Up to now 
MathSpeak had been seen as yet another speech standard by me, but it 
does seem like Dr Nemeth did put a lot of effort into making it work 
well with other things (eg. his Braille code).

Michael Whapples
On 06/10/2013 17:30, Susan Jolly wrote:
> Thanks, Kevin, for pointing out this video.  I watched (mostly 
> listened) to the whole thing although I stopped and started quite a 
> few times as I found it pretty boring.
>
> If I'm understanding correctly, Chrome is an operating system 
> developed by Google and ChromeVox is their competitor for Apple's 
> VoiceOver screenreader. There is also a Chrome browser which can be 
> installed on most other operating systems and ChromeVox can then be 
> used on these operating systems as an extension to the Chrome browser.
>
> The latest version of ChromeVox can speak math represented by either 
> presentation MathML or by alt tags associated with pictures of math.  
> Since it is not always possible to infer the semantics of presentation 
> MathML, it is not necessarily an optimal electronic representation of 
> math intended to be spoken.  The ChromeVox developers have addressed 
> this issue by providing an API that lets MathML content developers 
> annotate their MathML expressions with semantic enhancements intended 
> to make the corresponding math spoken by ChromeVox more natural and 
> also to make the expressions easier to navigate. For example, a 
> developer can choose to have the numerator of a fraction spoken using 
> a different pitch from its denominator.
>
> (It wasn't possible for me to tell from the video how the API actually 
> works and I haven't been able to find any documentation.)
>
> I was not impressed.  In the first place, I don't think it is 
> reasonable to expect authors to provide non-standard annotation that 
> is only recognized by a particular browser. In the second place, none 
> of the ideas for making spoken math easier to understand seem to be 
> new.  Researchers have long proposed various schemes that use prosody 
> for this purpose.  It is my opinion that none of these have caught on 
> because understanding spoken math is intrinsically difficult.  I 
> realize that there are many people for whom this is the only option 
> and the fact that the user can apply some level of customization to 
> ChromeVox spoken math may turn out to be valuable.
>
> Another purpose of spoken math is as a way of dictating math to a 
> person or app that converts the math to written form.  Dr. Nemeth's 
> MathSpeak was designed for this purpose and researchers at ghBraille 
> demonstrated a few years back that persons listening to MathSpeak are 
> less likely to misunderstand what is being said than persons listening 
> to other forms of spoken math.  Dr. Nemeth found that it typically 
> took no more than 15 minutes for him to train a sighted person with no 
> background in math to read all levels of math to him using MathSpeak.  
> Since MathSpeak is essentially a spoken form of Nemeth braille math, 
> Dr. Nemeth was able to braille what was being read to him as it was 
> being read.
>
> Nemeth braille math is effectively a very efficient shorthand for 
> entering a large portion of presentation MathML and I think that is 
> one of its many advantages over other alternatives for writing and 
> reading math in braille.
>
> Sincerely,
> SusanJ
>
> _______________________________________________
> Blindmath mailing list
> Blindmath at nfbnet.org
> http://nfbnet.org/mailman/listinfo/blindmath_nfbnet.org
> To unsubscribe, change your list options or get your account info for 
> Blindmath:
> http://nfbnet.org/mailman/options/blindmath_nfbnet.org/mwhapples%40aim.com 
>