What I did on my summer holidays.

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Geoff, you're coming at the same issue that Victor has been asking about, 
just from a different direction. In point of fact, you are both wrong. 
Adding smil data after the fact has proven very time consuming. In the 
case of reissues of old analog recordings, it is even more difficult 
because the original source print books are often unavailable in their 
same editions. Of course, it is possible to conceive that what you're 
thinking of could work, and it will once the speech recognition systems 
become more robust--but that's a diagression, just now.

The proper production sequence is something like the following:

1.)	Markup the text to the DTD, proof and validate;

2.)	Record the audio using live recording tools that also support 
marking as you go. In other words, the marked up text prepared in Step #1 
above is used onscreen (or on refreshable braille display) as the script 
for the narrator. Either the narrator, or the quality control person punch 
a button at every mark point--easy enough at the paragraph level, tedious 
for sentences, and impossible at the word level. For word level, we'll 
need the reco tools;

3.)	Proof and correct;

4.)	Generate distribution media from archive masters and ship;

OK, now about the speech reco. It's different from what is usually meant 
by the term because this time the computer knows in advance exactly what 
phonemes to expect and in what order--because it has the text in advance. 
So, rather than recognizing a word out of the universe of all possible 
words, it need only find the onset and termination of each word in its 
text file.

 On Tue, 
29 Jan 2002, Geoff Shang wrote:

> On Mon, 28 Jan 2002, Janina Sajka wrote:
> 
> > There is a profound difference between recording digitally and the DAISY
> > standard. If you only record, from beginning to end, you're functionally
> > no different than the analog cassette. Instead, DAISY imposes hierarchical
> > structure onto the recording, using the SMIL protocol. That way, you can
> > "rewind" and "fast forward" to something meaningful, because it's
> > structural, unlike today's media which only "rewind" or "fast forward"
> > some number of inches of tape irrespective of the actual intellectual
> > contents.
> 
> Yeah I realise this, but having it digital to begin with is going to save
> you the time needed to import it from analogue.  Not to mention that
> quality is likely to be better - if they've invested in digital, chances
> are they've invested in some decent recording equipment and environment.
> 
> Geoff.
> 
> 
> 
> _______________________________________________
> Speakup mailing list
> Speakup at braille.uwo.ca
> http://speech.braille.uwo.ca/mailman/listinfo/speakup
> 

-- 
	
				Janina Sajka, Director
				Technology Research and Development
				Governmental Relations Group
				American Foundation for the Blind (AFB)

Email: janina at afb.net		Phone: (202) 408-8175

Chair, Accessibility SIG
Open Electronic Book Forum (OEBF)
http://www.openebook.org





[Index of Archives]     [Linux for the Blind]     [Fedora Discussioin]     [Linux Kernel]     [Yosemite News]     [Big List of Linux Books]
  Powered by Linux