🚀 go-pugleaf

In article <1992Apr8.150518.962@agog.com> greg@agog.com (Greg Cockroft) writes:

   > 3. It would be nice if Simon would playback my voice commands so I could
   > remind myself of how I enunciated them. I sit there sometimes repeating
   > commands with permutations and combinations of inflections to the point  
   of
   > giving up and retraining them.
   > 

   We have had several people ask for this. The sounds are purposely deleted  
   to save disk space. If a word is approx. a second and sampled 3 times.  
   This is 24K a word. If you have 100 words this is 2.4 Meg of disk space.

Save only the version of the word that best matches the stored correlation
"template", and downsample it to 5K.  The playback will be telephone
quality, but it will convey the intonations the user is looking for.
And now you're down to 500K of disk.

--
Hugh Secker-Walker                    INTERNET:  hugh@ear-ache.mit.edu
MIT, Research Lab of Electronics      UUCP:      mit-eddie!mit-athena!hugh
     "The good news is the ear has 30,000 channels to the brain,
      the bad news is they're all nonlinear" -- M.B. Sachs