[Blindmath] Some examples of mathematics speech text

Fri Apr 15 21:03:02 UTC 2011

Using prosody has been shown to be quite useful.  MathPlayer has used
it for years and can generate SAPI 4,5 and SSML prosody tags.  Here's
the rub:  while some AT software makes use of this, the leading screen
readers do not and MathPlayer is forced to do its best by adding
commas and periods to introduce pauses.  If you self voice, then you
won't have a seamless reading experience unless you take over reading
the text as well... and then you are doing what a screen reader does.
Firevox took this later approach (sadly, Firevox seems like a dead
project).

There's a second rub: even when a voice claims support for SAPI or
SSML, quite often they only support some prosody features and are just
broken on others.  You test your system on one or two voices and it
seems good, and then try and third and there are big problems.  There
is a ScanSoft voice that that I know of that moves pauses around that
are embedded in the speech stream and simply refuses to speak anything
if there are two adjacent pauses.  That's maybe more extreme than
normal, but that's what the real world is like and limits what you can
do.  Eg, if you try your system on a voice that supports sounds, what
will it sound like to people who use voices that don't support sounds?
 Donal Fitzpatrick's work is interesting, but getting it to work
seamlessly with screen readers is problematic, at least at the
present.  SSML supports audio, but not SAPI.  The vendors that I have
spoken to only use SAPI voices.

Neil Soiffer
Senior Scientist
Design Science, Inc.
www.dessci.com
~ Makers of MathType, MathFlow, MathPlayer, MathDaisy, Equation Editor ~

On Fri, Apr 15, 2011 at 11:57 AM, Michael Whapples <mwhapples at aim.com> wrote:
> Hello,
> I agree with John on the idea that may be for single elements ends don't
> need marking, like how LaTeX doesn't require the braces. If speaking an
> equation under such a condition I probably would leave a slight pause just
> after the element to indicate this.
>
> Now to something I was going to mention when you were asking about the
> speaking of an equation, however this probably only works in the actual
> spoken form rather than the written down "what should be spoken". Consider
> how prosody might be used. Here is an example (I am only using pauses but
> speed may be good as well):
> 1 over x <pause> plus 2 x squared
> or,
> 1 over x plus 2 <pause> x squared
> or,
> 1 over x plus 2 x <pause> squared
> Probably for the above brackets may make things clearer but hopefully it
> shows how pauses could be used to make things clearer.
>
> If using a text to speech synthesiser the above can be done using SSML which
> can be passed to the synthesiser (espeak is one which accepts SSML, there
> are others). As part of the MSOR work I did I did implement this.
>
> This idea can be broadened to use more than just speech, possibly positional
> audio (eg. bracket being said from the left could indicate open bracket
> saving on a word, similar for fractions, etc). Also how about speeding these
> key words up very fast and/or shortening them where it seems sensible (eg.
> saying frac instead of fraction)?
>
> Instead of continuing with this I will point you to some research which I
> saw last year at the International Conference for Computers Helping People
> with Special Needs (ICCHP), you should have access to it through the Open
> University library system, its on Springer. The paper is called:
> "Spoken MathematicsUusingPprosody: Earcons and Spearcons"and I believe is by
> D. Fitzpatrick and E. Bates, published as part of the ICCHP 2010
> proceedings.
>
> I have to say I was very impressed with the idea and think it has great
> potential. Sometimes I think these formal spoken systems seem so clunky.
>
> Michael Whapples
>
> On -10/01/37 20:59, John Gardner wrote:
>>
>> Hello Jonathan, I like your philosophy of writing equations.  I can
>> suggest
>> two big improvements.  One is to compact the vocabulary so that you write
>> things like end fraction as EndFraction.  This helps a lot when one is
>> stepping through the equation with CTRL right-arrow and is fine for audio
>> recordings too.  For text that is to be read by a screen reader, adopt an
>> extension of the Latex rule that drops braces for single character sub and
>> superscripts.  I do not require the equivalent of braces when sub or
>> superscripts are written as a single string.
>> Let me illustrate. First your examples.
>>
>> Sum of two fractions: fraction 2 over 3 end fraction minus fraction 1 over
>> 6
>> end fraction
>> I would write: fraction 2 over 3 EndFraction minus fraction 1 over 6
>> EndFraction.
>> This reads a bit easier/faster when CTRL right-arrowing through, and it
>> helps even more when there are lots of end things.  Screen reader defaults
>> pronounce "split" words like EndFraction as end fraction and not
>> endfraction.
>>
>> Number to negative fractional power: 125 sup minus fraction 1 over 3 end
>> fraction end sup
>> I would write: 125 super -fraction 1 over 3 EndFraction
>> The superscript is written as a continuous string so doesn't require an
>> EndSuperscript.
>>
>> Numerator of quadratic formula. minus b plus or minus square root of b
>> squared minus 4 a c end root
>> I would write: -b PlusMinus SquareRoot b squared -4ac EndRoot
>> This reads fine except that the -4ac says negative 4 ack with my screen
>> reader, but in stepping through I stop at this term and arrow through to
>> hear it say dash 4 a c.  Not good for an audio recording, but fine for me.
>>
>> Gravitational attraction. cap G fraction m sub 1 m sub 2 over r squared
>> end
>> fraction
>> I would write: F sub G = G fraction m sub 1 m sub 2 over r squared
>> EndFraction
>> I find Cap G to be annoying, and the case of characters is usually clear
>> from context.  It's easy enough for me to arrow to the G and hear that it
>> is
>> a capital.  There are others who really want to hear that cap, so you need
>> to give users the ability to control how such things are spoken.
>>
>> Concentration of a solution: 10 sup -3 end sup micro grammes per
>> millilitre
>> I would write: 10 super -3 micro grammes per milliliter
>> Don't need the EndSuper because the -3 is written as a single string.  By
>> the way, it's often hard to tell the difference between sup and sub, so I
>> write super and sub.
>>
>> A few more examples might be instructive.  The full quadratic equation
>> solution:
>> x = Fraction -b PlusMinus SquareRoot b Squared -4ac EndRoot over 2a
>> EndFraction.
>>
>> Let me give an example of some difficult notation that gets really messy
>> normally.  A common notation in nuclear physics uses an element symbol
>> with
>> both left and right subscripts and superscripts. For example I once used
>> the
>> 111-indium metastable isotope whose full notation  can be written in Latex
>> as
>> _{49}^{111}In_{62}^{m*}
>> [Note that the sub and superscript order is not important - the
>> superscript
>> can be first and the subscript second.]  I would write this expression as:
>> LeftSubscript 49 LeftSuperscript 111 In subscript 62 superscript m*
>> which is spoken perfectly except for the In, which would need to be
>> stepped
>> through to hear that it is I n.  [Yes I know this is not quite correct,
>> the
>> In should be expressed in roman font]  For comparison, my MathPlayer
>> speaks
>> this expression as:
>>  sub 49 also super 111,  cap i, sub-superscripted n sub 62 super m  times,
>> end sub-superscripted
>> This and the Latex expression are, I claim, a whole lot harder to
>> understand
>> than my compact notation.  By the way, the MathPlayer expression is wrong,
>> because it somehow makes the "n" In "In" a subscript.  I've asked Neil to
>> check this out to find out whether the mistake is mine or MathPlayer's.
>>
>> What about more complicated sub and superscripts that need an EndSub or
>> EndSuperscript?  For illustration I distinguish the two cases:
>> x^n_i and x^{n_i}
>> respectively as
>> x superscript n sub i
>> and x LongSuperscript n sub i EndSuperscript
>>
>> Okay, that's my opinion.  By the way, Roopakshi mentioned Aster.  It uses
>> prefix notation.  Prefix notation is used in Content MathML and to some
>> extent in Presentation MathML.  For example, the (infix) expression x sub
>> I
>> is, in prefix notation something like sub x i.  Good for computers, and TV
>> Raman would argue that you and I should use prefix notation too.  As I
>> said,
>> there is no single "right" way to express math verbally.
>>
>> I look forward to some interesting commentary.
>>
>> John
>>
>>
>>
>>
>
>
> _______________________________________________
> Blindmath mailing list
> Blindmath at nfbnet.org
> http://www.nfbnet.org/mailman/listinfo/blindmath_nfbnet.org
> To unsubscribe, change your list options or get your account info for
> Blindmath:
> http://www.nfbnet.org/mailman/options/blindmath_nfbnet.org/neils%40dessci.com
>