When Response Matches Field
The "When Response Matches" field can be a phrase, a phrase template, or a specialized response template. If the match field is a word, phrase, or phrase template, the caller will need to speak a word or phrase to generate a match. For spoken input, the caller must wait until the prompt has finished before speaking; touch tones can be entered at any time. Improve speech recognition accuracy by entering complex words in phonetic forms: for example "straw berries" instead of "strawberries", or "Hue stun" instead of "Houston".
Note: The user can press the * key during speech recognition sessions to cancel the session and go back to the beginning of the Call Tree.
The match field can also be any one of several specialized response templates for common items (<TouchTone>, <Date>, <Time>, <PhoneNumber>, <SpelledWord>, etc.). You can access specialized templates by clicking the dropdown list for this field. If you select a specialty template, only one response will be allowed for this Node: all the conditional branching that results from the response should be handled at the next Node or in the VBScript Macro.
For examples that show how to accept different types of responses, look at the section on Examples of Accepting Different Responses.
Specialized Response Templates
Only specific specialized response templates can be used to accept touch tones. Some specialized response templates also accept spoken input.
The specialized response templates are defined as follows:
- <TouchTone>: Accepts the values 0-9, *, and #. The user can enter one touch tone via their telephone. This is one of the most reliable ways to accept data from a caller.
- <TouchTonesEndWith*#>: Accepts touch tones of up to 100 digits long that are terminated by * or #. The user can only enter touch tones via their telephone. The response will be set to the entire touch tone sequence entered by the caller, including the * or # terminating character. This is one of the most reliable ways to accept data from a caller.
- <Digits##>: Allows caller to enter a specific number of touch tone digits. To use the template, you must change the "##" to the number of digits you would like to input (include a leading zero if needed): for example, <Digits04> could be used for a PIN number, <Digits15> or <Digits16> for a credit card number. The response will be set to the entered digits. The tones # and * will be discarded. This is one of the most reliable ways to accept data from a caller.
- <PhoneNumber>: Accepts a US style area code and phone number that is either spoken or touch toned from the caller. The number can be either 7 digits or 10 digits, depending on if the area code is included. The response will be set to the phone number entered. The * key initiates as an abort feature that will return the caller to the beginning of the Call Tree. This template is far less reliable than either of the exclusively touch tone templates.
- <Date>: Allows the caller to speak a date (excluding the year). The template offers significant flexibility, for example - the caller could say any of the following: "Sunday", "Next Monday", "Tomorrow", "January 12", etc. The response will be set to a date in the form of "MM/DD/YYYY". The * key initiates as an abort feature that will return the caller to the beginning of the Call Tree.
- <Time>: Accepts a spoken time from the caller: "8:30 PM", "2:45 AM", etc. The template automatically clarifies AM/PM and returns the resulting time as the response. The * key initiates as an abort feature that will return the caller to the beginning of the Call Tree.
- <SpelledWord>: Accept a spelled word from the caller using NATO and police phonetic alphabets. Refer to the phonetic alphabets and special symbols page for information about accepted words and special characters. This template also provides special handling for two keywords "instructions" and "exit". The final response will be set to the spelled word or "**INSTRUCTIONS**" or "**EXIT**". The * key initiates as an abort feature that will return the caller to the beginning of the Call Tree.
- <YesNo>: Accepts a yes or no response as either a spoken response or touch tone: the user can press 1 for yes and touch tone 2 for no or speak "yes" or "no". Several spoken variations of "Yes" and "No" are understood (e.g., "Yeah"). The Node response variable will be set to either "Yes" or "No". The * key initiates as an abort feature that will return the caller to the beginning of the Call Tree.
Specialized response templates must be used by themselves and cannot be mixed with other words and phrases or phrase templates.
Phrase Templates
Phrase templates, which can be mixed with other words and phrases, are defined as shown below. The resulting responses always include the spoken input when the match field is a phrase template or a series text phrases. When using phrase templates, check the Response variable with message boxes to find which parts of the Response have to be filtered out (usually the phrase template is returned in the response).
- (JunkBegin): Any of a number of phrases that callers might say at the beginning of a sentence (e.g., "I want to" and "Please").
- (JunkEnd): Any of a number of phrases that callers might say at the end of a sentence (e.g., "please").
- [opt]: Use [opt] before a word or pre-defined template to mark it as optional. This will indicate to the speech recognition engine that the response can match regardless of whether the optional phrase is said. For example, "dollar [opt] sign" would allow the caller to speak "dollar" or "dollar sign" to generate a match. "[opt] hot dog" will allow the caller to speak "hot dog" or "dog" to generate a match.
- <Date>: Spoken dates.
- <Digit>: One spoken digit.
- <Digits>: A sequence of one or more digits.
- <Dollars>: Currency amounts.
- <Float>: Recognizes floating point numbers.
- <Fraction>: Basic fractions.
- <Integer>: Integers numbers (-12, 3, 5, -45, 0, etc.).
- <Month>: Months like "October" and "January".
- <Natural>: Natural numbers (1, 2, 3, etc.).
- <PluralNumber>: Plural numeric phrases (e.g., "1990's").
- <Time>: Spoken times.
Test Call Trees that use many phrase templates carefully, as speech recognition is often not as accurate as the user might expect or hope. Refer to the section on Limitations of the Software to learn more about why speech recognition on a telephone is extremely difficult.
Sample When Response Matches Values
Here are some sample "When response matches" field values and the corresponding responses that would be accepted:
|
|
When Response Matches field
|
Acceptable spoken responses that will generate a match
|
John
|
spoken "John"
|
<Digits16>
|
touch tone 16 digit number
|
<YesNo>
|
spoken: "yeah", "yes", "no" touch tone: 1 for yes, 2 for no
|
<Date>
|
spoken: "Monday", "Next Tuesday", "January 12", "Last Tuesday", "1 month ago", "3 months from today", etc.
|
<SpelledWord>
|
spoken example: "India Bravo Mary" translates to "IBM"
|
[opt] the President
|
spoken: "the President" or "President"
|
<Month> [opt] nineteen ninety nine
|
spoken: "January 99", "October 1999", etc.
|