Generic List Object

Performs Speech Synthesis and Speech Recognition. The speech recognition operates on both words and variables because the output ports provide variable substitution at run time. Teach me about variables.

Example: Tutorial6, Tutorial7, Tutorial8, and Tutorial9

 

 

 

 

Summary

Use the Generic List Object when you wish to provide branching based on speech recognition but are using variables instead of fixed words or phrases in the recognition ports. The variable in the recognizer is substituted when the dialog reaches the object.

The following diagram shows the recognition port of a Generic List Object. Speech recognition is performed on the words, phrases and variables within the text box. In this case the variable named "answer" is substituted and recognition is performed on both "the value of the variable called answer" and the fixed phrase "I canít remember."

In the Generic List Objectís speech recognition window, you may separate phrases just as you would with the Generic Object. (I.e. using a <return>) You may combine both words and variables in the recognition window

 

Details

  1. The variable $answer should be a string.
  2. Within the variable $answer, separate word groups using a comma (,)
  3. This is equivalent to separating words with a <return> in a recognition port.

    For example if the value of answer is "Animal, Vegetable, Mineral, Other", (note the commas)

    then the recognizer will spot any one of the words individually.

  4. Within the variable $answer, words separated by only a space are treated as a single phrase.

For example the value of answer is "Animal Vegetable Mineral Other" then the recognizer will only accept these words when uttered together as a phrase. (Note the absence of commas)

Specifying an exact pronunciation using phonemes

The Speech Toolkit uses the WorldBet symbols Rather than the International Phonetic Alphabet for phonetic representation. One reason is that the IPA has many characters that interfere with Tcl/TK and are not represented on standard keyboards. To help those familiar with IPA, there is an

IPA / WorldBet translator.

To insert phonemes into the recognizer using the IPA Keypad translator:

  1. Click the Insert IPA
  2.  

     

     

     

     

  3. Use the IPA buttons to build a word or phrase.
  1. Right Click the IPA symbols for an audio example ( or use audio button )
  2. Double Click the IPA symbols to insert the associated WorldBet phoneme in the Recognition Box ( or use insert button )
  3. Hold the mouse over the IPA symbol to view a brief Sticky Note description

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Type and Category

Type: Because the Generic List Object allows speech recognition on both words and phonemes, you must tell the system which one you are using. The default setting is "Words." Keep this setting if you are using words, phrases and variables that use standard letters. Use the optional "WorldBet" setting when inserting phonemes for recognition. You can not use both words and phonemes in the same recognition window.

Category:



The Speech Toolkit has an optional "Data Capture Mode" (see
Data Capture in the reference manual for details). The Data file is simply a tab delimited text file where each line represents information from an object during the dialog. Each Generic, Generic List Object, and PSL Response Object make an entry into the data capture file during a dialog. The Category menu allows you to classify a userís responses in the recognition port as either Pass, Trap, Help, Other or the default Undefined.

Pass: Use this setting to categorize the userís response as "correct" or "expected"

Trap: Used with *any in the recognition port. The userís response is incorrect or unexpected

Help: User says, "help" or requests assistance in the dialog

Other: Reserved for your own classification

Undefined: Category not specified

 

 

 

 

 

 

 

 

 

 

 

 

 

 


Properties (Generic List Object)

Each objectís preferences window has a subset of these tabs.

 

 

TTS Edit the text to speech

Recorded Edit recorded speech or pre-process TTS prompts

Recognition Edit speech recognition settings, change recognizers, repair dialogs.

DTMF Edit touch tone telephony settings

Misc Edit miscellaneous settings

On Enter Create a Tcl/TK script to be run when the dialog enters the object

On Enter Create a Tcl/TK script to be run when the dialog exits the object

Tucker-Maxon Edit Tucker-Maxon Package settings

Use Select TTS (text to speech) or recorded voice

Face Select from available user defined and default face configurations


Generic List Object Options

Properties

Opens the properties window for the action box

Add Port

Creates an additional RED Speech Recognition Port for branching

Cut

Deletes the object and copies it to the clip board

Copy

Copies the object to the clip board

Delete

Deletes the object

Rename

Renames the object as it appears on the canvas