home *** CD-ROM | disk | FTP | other *** search
- <html>
-
- <head>
- <meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
- <meta name="GENERATOR" content="Microsoft FrontPage 3.0">
- <title>How does the speech interface work for the programmer ?</title>
- </head>
-
- <body bgcolor="#FFFFFF">
-
- <p><font size="6" color="#FF0000">How does the speech interface work for the programmer ?</font><br>
- </p>
-
- <hr>
-
- <p><br>
- <font face="Verdana">The following interface is currently available and recommended...<br>
- <br>
- 1) Recording starts as per the settings for frequency.(USE 22050 or just ignore the
- instruction)<br>
- <br>
- 2) We/You announce RECORDING start by a wave file play. Operation Status displays
- "Alert"<br>
- <br>
- 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<br>
- <br>
- 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.<br>
- Also this is natural and human can easily adopt to repeating a word as we often do it
- during our conversations.<br>
- <br>
- 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.<br>
- 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 <br>
- bit longer time.<br>
- <br>
- 6) After the processing is over the result is available through various ways like
- WORDS_I_WANT or LANGUAGE EDITOR Platinum word.<br>
- <br>
- 7) Now the Operation Status displays "Finish", again a wave announcement of the
- file in path to Play_this_file_Start is made.<br>
- 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". <br>
- SEE all of our examples.<br>
- <br>
- 8) The Operation Staus again displays "Alert" and VoiceAction is ready to hear
- your next word<br>
- <br>
- 9) ALL THIS ( 1-7 STEPs above)usually consumes ONE SECOND or TWO SECONDS on a 350 MHz
- Machine<br>
- Hence we recommend you this interface which is extremely innovative, attractive ,genuinely
- interactive and also consumes less <br>
- of all and everything. <strong><em><big>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?.</big></em></strong><br>
- <br>
- If you still feel to have a different kind of interface just e-mail to
- urlabs@pn2.vsnl.net.in <br>
- we will give you as you desire it.<br>
- <br>
- 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<br>
- to a different one. How about a guitar chord or a instrumental ting?</font></p>
- </body>
- </html>
-