Speaking text and highlighting speaking text

[Ben Rubinstein]

on 4/3/02 6:07 PM, Glasgow, David at David.Glasgow at cstone-tr.nwest.nhs.uk
wrote:
> 1/ I like to have my Mac read big blocks of statistical data [snip] is this
> hard to implement under Windows?  Is there a WinVict dot something?  My PCs
> have never uttered a word to me...... I can't seem to find references to
> synthesised speech, but do recall it being mentioned previously on this list
> (or maybe MC)

I believe that there is an external, in the famous "externals collection"
that was available on the original Revolution web page, that provided
_cross-platform_ access to text-to-speech.  But I've not used it.

> 2/ In some ways related, perhaps.  I have a psychological assessment that
> speaks the questions for people with a reading difficulty.  It speaks once
> automatically, and any additional number of times at the users request.  I
> highlight the whole sentence by colouring the text as the thing is read, and
> then it turns to black again.  Now, how would I highlight *words* as they are
> being read.  Like you get in Grandma and Me?  I have assumed that chopping up
> the Aiffs into words would be a silly way to go.  So perhaps I use QuickTime
> and colour/decolour words at particular points in the playback?  I know I
> could just experiment, but I am confident that someone out there must have
> tackled this previously.

I'm not clear whether you intend this to work with text-to-speech, or with a
pregenerated recording of someone actually reading the text.  If the latter,
while there are obviously many interesting options, one of the more obscure
but potentially best (and arguably most fun) is that QuickTime has a little
known 'karaoke' facility.  This works cross-platform - essentially some
additional data is encoded in the text track, which tells when to colour
parts of each word.  It was added five or six years ago when karaoke was
popular in some quarters, and a defacto standard grew up for a ".kar" file
format, for which lots of people produced players, and (the good bit) also a
variety of Windows and Mac apps to assist in the production of these files.
Since QuickTime gained the ability to import this format, these apps can be
used to generate this data, which is by far the easiest way.  I think it's
designed to work with Midi, but I'm sure you could create a null Midi track,
and actually sequence it with digitised audio.

To see what it looks like, just search for "QuickTime Karaoke" in Google -
you'll quickly come across loads of samples.

Important disclaimer: I've never sung karaoke in my life.  There are good
historical reasons why I know this stuff.  It's not my fault.  Honest.
 
  Ben Rubinstein               |  Email: benr_mc at cogapp.com
  Cognitive Applications Ltd   |  Phone: +44 (0)1273-821600
  http://www.cogapp.com        |  Fax  : +44 (0)1273-728866