WIXLIB

International Journal of Computer Applications (0975 – 8887)Volume 74– No.14, July 2013A DNS lookup service is sometimes also required to resolvethe named address. Multi-threading safe third party librariessuch as libcurl etc. are used to achieve high performancedocument fetching in the engine.encounters prompt element or block element with a text, itneeds to be played as an audio to the caller [10]. The VXMLinterpreter sends this audio text or a .wav file location to themedia resource which in turn plays it to the caller.4.1.2.2 Form Interpretation Module4.3 Telephony ResourceThe form Interpretation Module is responsible for theimplementing the functional logic and traversing through theVXML document based on caller inputs [7]. The forminterpretation algorithm (FIA) drives the interaction betweenthe caller and a VoiceXML input items. Execution of the codein forms is handled by the form interpretation algorithm(FIA), which loops through the items in the form andprocesses (or reprocesses) them.The telephony resources include DTMF inputs and Callcontrol. DTMF inputs serve as caller responses to the Voicegateway. The call control features such as call disconnect, calltransfer, conference etc. which are sent from the telephone arehandled by the telephony resource. In order to achieve thesefunctionalities Call Control Interpreter and VXML Interpretershould work in conjunction as described in the fig. 3.According to the W3C recommendation, FIA must handle thefollowing [7] Form initialization. Prompting or delivering audio to the caller. Grammar activation and deactivation Entering the form with an utterance that matchedone of the form's document-scoped grammars whilethe user was visiting a different form or menu. Leaving the form because the user matched anotherform, menu, or link's document-scoped grammar. Processing multiple field fills from one utterance,including the execution of the relevant filled actions. Selecting the next form item to visit, and thenprocessing that form item. 5. CCXML GATEWAYThe CCXML (Call Control XML) gateway is required tohandle call control requirements that are beyond the scope ofthe VoiceXML specification. Although the CCXML andVXML can be used in conjunction, both are mutuallyindependent. VoiceXML does support certain call controlfeatures like transfer etc. but these would not be enough forproviding complex solutions for call controls, which would berequired in IVR systems. Hence, there are two approaches indesigning an IVR system using the VXML. The first approachcombines SIP as a control language and VoiceXML as theinteraction language [11]. The second approach is to useCCXML as a control language. Former being more used withtraditional IVR systems built with CCXML and the latter isused in more recent and advanced IVR systems. CCXMLoffers more controlled and advanced call control features withthe advantage of short development time required toimplement and modify the system. The followingrequirements are addressed by this w3c specification [12]: Choosing the correct catch element to handle anyevents thrown while processing a form item.The interpretation algorithm of FIA is broken into threedifferent phases viz. select, collect and process phase [8]. Inselect phase the VXML interpreter selects the form item(input item) which needs to be processed. Then in the collectphase it collects the caller input against the selected item andvalidates with the active grammars which are basicallyvalidation rules for the input item [9]. Finally in the processphase, it processes the user input as per the execution logicdefined in the VXML document. 4.1.2.3 XML Parser ModuleThe XML Parser module acts as the interface between theengine and the third party XML parsing and validation library.It is also responsible for handling the tags pertaining to theparsed DOM trees. It also contains functionality which allowsthe user to validate dynamically generated XML scriptsagainst DTD and Schema. Its major functionality is to convertthe parsed VXML file into a binary format to facilitate quickcycles for the Form Interpretation Module.4.2 Media ResourceThe media resource is responsible for speech recognition andgrammar validations for the input received from the caller asan audio. It matches the input against the active grammars andconverts the audio response to binary format which could beunderstood by the VXML Interpreter. It is also responsible forconverting text to speech and audio playback. Whenever FIA Support for multi-party conferencing, with advancedconference and audio control. A conferencing applicationinvolves multiple participants, and is dependent upon callcontrol to establish relationships between thoseparticipants.The ability to give each active call leg its own dedicatedVoiceXML interpreter. For example, in VoiceXML, thesecond leg of a transferred call lacks a VoiceXMLinterpreter of its own, limiting the scope of possibleapplications.Sophisticated multiple-call handling and control,including the ability to place outgoing calls.Handling for a richer class of asynchronous events.Advanced telephony operations involve substantialamounts of signals, status events, and message-passing.VoiceXML 2.0 does not integrate asynchronous"external" events into its event-processing model.VoiceXML lacks the external interfaces required tointeract with an outside call queue, or place calls onbehalf of an external document server.The Voice XML when used with CCXML provides thebenefits such as lower code complexity and higherextensibility [11]. In such a scenario, the engine and gatewayonce developed does not require repeated and frequentmodifications. The only customization that is required is increating or modifying the VXML/CCXML documents whichare easier to code when compared to traditional codedeveloped in a high level language. Hench the codecomplexity is reduced. Also XML based languages are farmore extensible when compared with SIP based architecture.33

International Journal of Computer Applications (0975 – 8887)Volume 74– No.14, July rolInterfaceMediaDialogServerCallerFig 3: CCXML – VXML Integration Architecture6. DEVELOPMENT TOOLS There are numerous resources which can be used to quicklycreate or modify an error free VXML document. Followingare few of them:6.1 Plain text IDEThese are integrated development environments for creatingand modifying a VXML/CCXML document. The IDEsusually should have any or all of the following features: Color Coding – This enables the developer toquickly identify the reserve keywords, matchingtags, custom strings for prompt and audio playbacksetc.Inline syntax checking – The IDEs usually act like amini interpreter; in which code can be continuouslyparsed while it is being edited, providing instantfeedback when syntax errors are introduced.Document traversal – The VXML documents maycontain certain transition tags which require a newdocument to be loaded. These IDEs are also able toimmediately load the next document as soon as theuser clicks on the hyperlink or the URL.6.2 GUI based IDEUsing the GUI based integrated development environmentsdeveloper can use the drag and drop feature to create ormodify any VXML document. These IDEs serve the samepurpose as Plain Text IDEs but differ in the following: Drag and Drop – If a developer wants to create ormodify any tag in the VXML document, it is notrequired to type the tag and its parameters. Instead,the developer drags the required tag from theavailable tag tools and it is automatically populatedat the desired location.Proactive rather than reactive – The IDE does notallow developer to insert any incorrect tags orcorrect tags at incorrect locations. Hence, it actsproactively rather than throwing an error after thedeveloper finishes typing.Large binary size –GUI based IDEs are complex todesign and have a very large binary or source codesize.Low availability – Since these types of IDEs requirea lot of monetary and development effort, these arenot easily available commercially as well as in theopen source domain.7. PERFORMANCE MEASUREMENTAND OPTIMIZATIONPerformance is one of the key features of any VXML orCCXML interpreter. It is measured as how the integratedsystem performs in terms of responsiveness and stabilityunder a particular workload. The approach suggested heretreats the complete gateway as a black box. Artificial load isgenerated from a different program which is known as aworkload generator. The workload generator calls the APIsexposed by the telephony interface. As soon as a call isreceived on the telephony interface through the workload34

International Journal of Computer Applications (0975 – 8887)Volume 74– No.14, July 2013generator, it follows the normal execution path to completethe call which would typically involve session initiation,fetching documents, parsing, traversal and dialoginterpretations etc. When it completes the call it returns to theworkload generator with the status of the call. This statuswould comprise of time taken, success/failure etc. A numberof calls are generated from the workload generated in order toperform a load test on the system. The through put of thesystem varies depending on many reasons but the followingtwo are the critical ones to look at:𝑇ℎ𝑟𝑜𝑢ℎ𝑝𝑢𝑡 1number of calls to the interface𝑇ℎ𝑟𝑜𝑢ℎ𝑝𝑢𝑡 1size of the VXML or CCXML documentThe first one is more of a business requirement that drives theload testing and cannot be regulated or improved throughtechnical improvements in the system. The second one ismore relevant for making improvements to the system. It isobserved that the performance degradation for hugedocuments is due to two major factors: Time taken in downloading the document.Parsing the document.Caching could be used in order to reduce the time taken todownload the document. The document should be cached andchecked before downloading, for any modifications that havebeen done on the previously downloaded document. If thedocument has not been modified from the last download it canbe fetched from the cache. Similarly, a caching mechanismcan be developed for the parsed binary tree which is createdafter the parsing of the document as described in section4.1.2.3 of this article. These two techniques could be used inconjunction with each other in order to make majorimprovements in the performance of the VXML Engine.8. PLATFORM CERTIFICATIONIn order to establish a VXML based IVR system, the platformhas to go through a series of validation tests which areprovided by the VXML forum [13]. These tests are based onW3C VXML specifications and standards. These are basicallyan exhaustive set of tests which act as a benchmark for theplatform based on the W3C standards. The VoiceXMLForum’s Platform Certification Program usually providesvendor-independent, industry-standard certification thatsupports all parts of the VoiceXML ecosystem.IJCATM : www.ijcaonline.org9. ACKNOWLEDGMENTSThe author would like to express his sincere thanks to all thepeople in the VXML Interpreter development team for theirhelp and support without which it would not have beenpossible to get the deep insights in the Voice XML.10. REFERENCES[1] From Wikipedia, description about interactive voiceresponse systemhttp://en.wikipedia.org/wiki/Interactive voice response[2] VoiceXML Forum is a global industry organization thatworks to accelerate the adoption of VoiceXML andadjacent technologies. The reference is taken from thefrequently asked questions of the ed-questions[3] Anderson, E. A., Breitenbach, S., Burd, T.,Chidambaram, N., Houle, P., D. Newsome, D, Tang, X.,Zhu, X., Early Adopter VoiceXML, Wrox, 2001 p 24.[4] W3C Recommendation for VoiceXML 3.0.http://www.w3.org/TR/voicexml30/[5] Adam Hocek, David Cuddihy, Prentice Hall ProfessionalPublished: January 2003, Definitive VoiceXML[6] The Staff of DreamTech Inc, McGraw-Hill Companies2002, VoiceXML 2.0 Developer's Guide[7] W3C Recommendation for VoiceXML 2.0.http://www.w3.org/TR/voicexml20/[8] A. Larson, Prentice Hall Professional TechnicalReference 2002, VoiceXML: Introduction to DevelopingSpeech Applications[9] W3C Recommendation for SRGS grammar.http://www.w3.org/TR/speech-grammar/[10] Bob C. Edgar, C M P Books, 2001, the VoiceXMLHandbook.[11] Daniel Amyot and Renato Simoes 2007 “CombiningVoice XML with CCXML: A Comparative Study”,Consumer Communications and Networking Conference,2007.[12] W3C Recommendation for CCXML 1.0.http://www.w3.org/TR/ccxml/[13] VoiceXML Forum is a global industry organization thatworks to accelerate the adoption of VoiceXML andadjacent technologies. The reference is taken from theplatform certification section of the on35

4. VOICE XML GATEWAY The components included in the voice xml gateway revolve around the engine which is known as the VXML interpreter. This engine is responsible for Mediating, Controlling and Distributing data and logic flow in and around the other components of the gateway. VoiceXML is not only used to