How does the speech interface work for the programmer ?
The following interface is currently available and recommended...
1) Recording starts as per the settings for frequency.(USE 22050 or just ignore the
instruction)
2) We/You announce RECORDING start by a wave file play. Operation Status displays
"Alert"
3) The user can say his word in between the kept Speech Sampling Time(Recording Time)If he
does'nt then VoiceAction Loops back to start again (but no wave announcement is made)This
is done to conserve memory and be extremely light weight on your program
4) What if speaks exactly on the edge of looping, well in that case, VoiceAction announces
the starting wave again. That is the machines 'Pardon?' Hard luck the user has to repeat
the word. This can be easily avoided by keeping a longer Speech Sampling time like above
20000.
Also this is natural and human can easily adopt to repeating a word as we often do it
during our conversations.
5) If he says in synchronization (serious users) then the speech IS captured and processed
as per the kept Sample Processing Resolution. Operation Status displays
"Processing...".More is the Sample Processing Resolution less is the processing
time less is the accuracy.Lesser the Sample Processing Resolution more the time high the
accuracy.
This entirely depends on the speed of your machine and also the type and number of words
you want to recognize. As example , when it comes to YES /No only keep SPR High. When it
comes to Edit, Delete keep it low 200 see example REMOTE INSTRUMENT CONTROL. Yes / No get
processed fast but for one two three it takes a
bit longer time.
6) After the processing is over the result is available through various ways like
WORDS_I_WANT or LANGUAGE EDITOR Platinum word.
7) Now the Operation Status displays "Finish", again a wave announcement of the
file in path to Play_this_file_Start is made.
Get the KEY idea..that is if you say /Hi/ and if the Platinum_Wordno1 shows "Hi"
too then you just have to change a name (or even path )to point to a different wave file
which says "You said Hi, O.K Hello user".
SEE all of our examples.
8) The Operation Staus again displays "Alert" and VoiceAction is ready to hear
your next word
9) ALL THIS ( 1-7 STEPs above)usually consumes ONE SECOND or TWO SECONDS on a 350 MHz
Machine
Hence we recommend you this interface which is extremely innovative, attractive ,genuinely
interactive and also consumes less
of all and everything. As the processors now are faster, its like you
speak and it writes and tells back what is going on and what you must do next . Good Isn't
it?.
If you still feel to have a different kind of interface just e-mail to
urlabs@pn2.vsnl.net.in
we will give you as you desire it.
Also due to the very nature of the Interface ,if you tend to get frustrated to hear
/Start/ after every word, well, use your common sense and remove the wave from the path or
change the wave
to a different one. How about a guitar chord or a instrumental ting?