Call Viewer: Call View page

Call View displays all available information about the selected call. Along with a summary of session-level information, the Call View page breaks down logged information about a call on a per state basis. This information is presented in tabular form. Each row of the table represents the logged occurrence of a state. A state may be any one of the following: Database state, Decision state, Play state, Dialog Module state, and Application-defined state.

IVRs produce two types of log data: Application logs and Recognition logs. There is no data overlap between these two types of logs. When Application logs and Recognition logs are both present, both the APPLICATION tab and the RECOGNITION tab are visible in the Call View page.

In the Call View page, to advance directly to the Call View of the next consecutive call in a Call List, click the Next ( ) arrow beside the session ID #. If you want to go back to the call immediately before the current one, click the Previous ( ) arrow .

Note: All three tabs of the Call View page use symbols to communicate various statuses, values, assigned categories, and rate of progress. Click here for more information about Call View's use of symbols.

APPLICATION tab

Application logs contain all information about a call except data associated with recognition. You can see a session's sequence of state occurrences, their respective state types, time stamps, durations, the task each state belongs to (if applicable), and the outcome (if the state is a Dialog Module).

If there are applicable dialog module tasks, Call View displays visual delimiters marking the start and end of each task. When required, to accommodate Dragon Voice interactions, Open Dialog Start and End delimiters mark the use of Open Dialog dialog modules.

RECOGNITION tab

What is displayed in the RECOGNITION tab depends on where the recognition data comes from. If incoming recognition data comes from Nuance Recognizer logs, Call View displays the Nbest ranking, meaning, recognized text and the numerical confidence value. If incoming recognition data are events from Dragon Voice components Krypton Recognition Engine and Natural Language Engine, Call View displays Intent/Concept, the numerical confidence value, grammar packages called upon, grammar-specified weights, and the Krypton-determined effective weights.

Data from Recognizer

Displayed data from Nuance Recognizer includes recognition results and their confidence scores as well as timing statistics associated with each recognition event.

Call Viewer view for Nuance Recognizer logs

Data from Dragon Voice

For recognition events from Dragon Voice components like Krypton Recognition Engine and Natural Language Engine, the RECOGNITION tab displays information based on Krypton and NLE events.

Call Viewer view for Dragon Voice events

For more details about Call View's display of Dragon Voice recognition data,

click here.

Displayed parameter	Description
Prompt	This is the question or prompt asked of the caller. The value comes from the `NVOCinpt` event and is stored in the `TEXT` token. If you mouse over the prompt, you will see the URL for the audio file of this prompt in a tooltip. Note: This event is a Nuance Vocalizer event and not a Dragon Voice component event. This event is applicable to Nuance Recognizer data as well as Dragon Voice data. The following is a segment of a line from a recognition log in which the prompt is defined: `>...` `\|EVNT=NVOCinpt\|``MIME=application/synthesis+ssml\|TXSZ=295``\|TEXT``=<?xml version\="1.0"?><speak xml:lang\="en-US" version\="1.0" xmlns\="http://www.w3.org/2001/10/synthesis"><audio fetchhint\="prefetch" fetchtimeout\="5000ms" src\="http://10.20.19.26:8080/PayBill/en-US/prompts/application/default/pb8060_ini_01.wav?version\=xx.yy.zz">``From which account?``</audio></speak>` Note: The above example is a subset of information contained in a full recognition event line, shortened to emphasize one or more elements.
Speech input	This is the recognized text from what the caller said. The value comes from the `KRrslt` event and is stored in the `RMFF(0)` token (the `0` indicating the highest confidence result). `...` `\|EVNT=KRrslt\|``BORT=6403\|DLMOOV=http://10.20.19.26:8080/PayBill/en-US/ ... KRCONF(7)=0\|MEDIA=audio/basic;rate=8000\|RCPU=14100``\|RMFF(0)=Honda pay two hundred dollars to visa on January twenty seventh twenty eighteen``\|RMFF(1)= ...` The example, above, highlights the Krypton event line for a given `KRrslt` event in the recognition log and the `RMFF(0)` token's value. Note: The above example is a subset of information contained in a full recognition event line, shortened to emphasize one or more elements.
Intent	This value comes from the `NLEplnd` NLE event and is stored in the `INTENT(0)` token. NLE logs this event just before converting the response to the Interpret command into the requested final format. It includes the final post-processed pipeline output. There is one such event for each Interpret command. The following is a segment of a line from a recognition log in which intent is defined: `...` `\|EVNT=NLEplnd\|INTENT(0)=BILL_PAY\|``KEYS(0)=<INTENT>BILL_PAY</INTENT><AMOUNT>200.00</AMOUNT><PAYEE>VISA</PAYEE><DATE>20180127</DATE><NLCONF>99</NLCONF><WHENCE>SSM</WHENCE>\|NBST=1\|NLCONF(0)=99\| ...` The example, above, highlights the NLE event line for a given `NLEplnd` event in the recognition log and the `INTENT(0)` token's value. Note: The above example is a subset of information contained in a full recognition event line, shortened to emphasize one or more elements.
Concept (in the Key column and is anything except Intent)	These values come from the `NLEplnd` NLE event and are stored in the `KEYS(0)` token as discretely named keys (for example `<DATE>20180127</DATE>`). In recognition logs, you may find there is more than one instance of a `KEYS` token, and instances are distinguished from one another with a numeral in parentheses. For example, there may be the tokens `KEYS(0)`, `KEYS(1)`, `KEYS(2)`, `KEYS(3)`, and so on, along with the keys that belong to each token. Having several token instances reflects the situation in which the recognition engine and interpreter return more than one possible result. The results are ranked in an Nbest list. Usually, `KEYS(0)` will contain the top result (highest confidence), `KEYS(1)` will contain the next best result, etc. The following is a segment of a line from a recognition log in which concepts are defined: `...` `\|EVNT=NLEplnd\|``INTENT(0)=BILL_PAY\|KEYS(0)=<INTENT>BILL_PAY</INTENT><AMOUNT>200.00</AMOUNT><PAYEE>VISA</PAYEE>``<DATE>20180127</DATE>``<NLCONF>99</NLCONF><WHENCE>SSM</WHENCE>\|NBST=1\|NLCONF(0)=99\| ...` The example, above, highlights the NLE event line for a given `NLEplnd` event in the recognition log and one of the concept keys from the `KEYS(0)` token. Note: The above example is a subset of information contained in a full recognition event line, shortened to emphasize one or more elements.
Confidence	This value comes from the `NLEplnd` NLE event is stored in the `NLCONF(0)` token. This represents the confidence level the NLE engine has in the assigned intent and is a value from 0 (zero) to 100. The following is a segment of a line from a recognition log in which confidence is defined: `...` `\|EVNT=NLEplnd\|``INTENT(0)=BILL_PAY\|KEYS(0)=<INTENT>BILL_PAY</INTENT><AMOUNT>200.00</AMOUNT><PAYEE>VISA</PAYEE><DATE>20180127</DATE><NLCONF>99</NLCONF><WHENCE>SSM</WHENCE>\|NBST=1\|``NLCONF(0)=99\|` `...` The example, above, highlights the NLE event line for a given `NLEplnd` event in the recognition log and the `NLCONF` key from the `KEYS(0)` token. Notes: NLE confidence values differ from Nuance Recognizer confidence values. Recognizer confidence is a value from 0 to 1000 while NLE confidence is from 0 to 100. The above example is a subset of information contained in a full recognition event line, shortened to emphasize one or more elements.
Grammars	These values come from the `KRrcst` Krypton event, and are stored in the `WEIGHTS` token in the case of grammars, and in the `WORDSETS` token in the case of wordsets (for example `\|WEIGHTS=http://10.20.19.26:8080/PayBill/en-US/grammars/application/PS_NDF_PayBill_KryptonModel.zip?nlptype=krypton&dlm_weight=0.7&version=xx.yy.zz`). There may be multiple items for DLMs and wordsets. In this event, which represents the Krypton recognition start, the runtime grammar references are specified via the ID provided in the corresponding load request (or created by the Krypton engine and returned in the load response). The following is a segment of a line from a recognition log in which grammars are defined: `...` `\|EVNT=KRrcst\|``LANG=eng-USA\|LANGVER=ndp-eng-USA-GEN-3.7.1\|MAXNONFACTORY=900\|MRECVER=1.39.100.18359\|S3VER=12.20.300.03803\|TPVER=15.04.100.01668``\|WEIGHTS=http://10.20.19.26:8080/PayBill/en-US/grammars/application/PS_NDF_PayBill_KryptonModel.zip?nlptype=krypton&dlm_weight=0.7&version=xx.yy.zz -1 -1 10000:699\|WORDSETS=http://10.20.19.26:8080/PayBill/en-US/grammars/application/FROM_ACCOUNT_wordset.json?nlptype=wordset&version=xx.yy.zz -1 -1 10000;http://10.20.19.26:8080/PayBill/en-US/grammars/application/PAYEE_wordset.json?nlptype=wordset&version=xx.yy.zz -1 -1 10000\|UCPU=0\|SCPU=0` The example, above, highlights the Krypton event line for a given `KRrcst` event and both the `WEIGHTS` and `WORDSETS` tokens. Note: The above example is a subset of information contained in a full recognition event line, shortened to emphasize one or more elements.
Weights	This value comes from the `KRrcst` Krypton event, and is stored in the `WEIGHTS` token as the `dlm_weight` parameter as a suffix for each grammar item in the event. Weight is a number representing the relative importance of each grammar package with respect to the others. Note: In the example for Grammars, above, neither of the two wordsets have a set `weight` value. When this happens, the system assigns the value `100` as a default.
Effective weights	The effective weight is a value calculated by the Krypton engine from the individual Weight values from DLMs and wordsets. Effective weight is the weight value used during recognition. In the recognition logs, this value comes from the `KRrslt` Krypton event and is stored in the `EFFWEIGHTS` token as a suffix parameter to the listed grammar packages. The following is a segment of a line from a recognition log in which effective weights are defined: ... `\|EVNT=KRrslt\|``BORT=3011\|DLMOOV=http:// ... UNICATIONS:0;VERT_TRAVEL:0\|``EFFWEIGHTS=http://10.20.19.26:8080/PayBill/en-US/grammars/application/PS_NDF_PayBill_KryptonModel.zip?nlptype=krypton&dlm_weight=0.7&version=xx.yy.zz -1 -1 10000:699\| ...` Note: The above example is a subset of information contained in a full recognition event line, shortened to emphasize one or more elements.
Recognition status (icon)	This icon represents the status of the call. The possible values for this are `SUCCESS`, `RECOGNITION-TIMEOUT`, `NO-INPUT-TIMEOUT`, `ERROR`, and `NO-MATCH`. If the color of the circle is green, the confidence level is 50% or above. If the color is red, the confidence level is below 50%. If the color is gray, the status is unknown or there is no input. `SUCCESS` - Indicates recognition successfully occurred with a high confidence. `RECOGNITION-TIMEOUT` - Indicates recognition did not successfully occur before the timeout period expired. `NO-INPUT-TIMEOUT` - Indicates recognition did not successfully occur before the input period expired. `ERROR` - Indicates something prevented a successful recognition. Examples of this are if Krypton cannot fetch a grammar, or if the audio passed to Krypton is in an incorrect format. `NO-MATCH` - Indicates recognition successfully occurred with a high confidence.
Barge-in	The presence of the lightning bolt within the status icon means the recorded speech was a barge-in which means it was an interruption of the prompt.

MERGED tab

The MERGED tab represents the merging of data from the Application logs and the Recognition logs. This tab is visible only if both Application logs and Recognizer logs are generated using a Nuance suite of products, in particular, supported versions of Nuance Development Framework (NDF) and Nuance Recognizer or Dragon Voice components like Krypton Recognition Engine and Natural Language Engine must be the source of Recognition logs.

Notes: