Services
A voice recognition-based application works by means of a man-machine dialog, where callers get to hear information and are initiated to provide the system with spoken input. The Voice User Interface (VUI) is the front-end of a voice recognition-based application. Its design must be carefully constructed in order to intelligently cover functionality and meet caller-expectations at the same time. Business models initially set the framework for any automation-project. As far as automation with voice recognition-based systems is concerned the VUI drives the ROI as it is mainly responsible for overall caller experience and acceptance.Because the success of a voice recognition-based application is highly depending on the quality of the VUI voiceandvision offers the following services:
Requirement Gathering and Analysis
- Customer Services (helpdesk, call center)
A logical starting point of any VUI design is a thorough analysis of customer service already in place (such as a help desk or call center). Already existing customer services provide necessary and crucial information in regard to the voice application to be created. Monitoring and analyzing dialogs with end-customers provide important input for the Sound & Feel Definition and Prompt Design.
With a VUI-design framework in mind voiceandvision professionally analyzes the interaction between service provider and end-customers in order to present a design strategy. - Functionality
Experience shows that the intended functionality that is to be covered by a future speech recognition-based system often is constructed out of internal development over time and marketing-driven initiatives.
In order to suit the intended functionality to a speech recognition-based application voiceandvision, within a VUI context, carefully defines the tasks of the application to be developed. - Infrastructure & Back-end
Defining functionality shows dependencies on back-end systems and process-related data-availability. Typical questions to be answered are: What comes back from the database? How can information be made available directly on the speech platform? What about back-end latencies and how to handle them? How to work with storage of large amounts of data, such as sound-files?
voiceandvision valuates and examines technical feasibility and infrastructure in order to propose processes that will realize an optimal call-experience. - Business requirements
Does the system need to handle an extensive call volume to make it profitable? Parameters like price per call or per minute set a clear target. Voice business is about finding profitable solutions and guaranteeing added value at the same time.
voiceandvision consults its customers in the development and analysis of business goals and models. - End-customers
To realize user-centred design it is crucial to know as much as possible about the caller. Typical question should address who is calling, why they’re calling, when they call, how they call, where do they call from, what do they want and what they know.
With dedicated call-monitoring, customer-tailored Wizard of Oz-testing and specially designed questionnaires voiceandvision describes end-user profiles in order to create user-centric VUI design. - Marketing & Corporate Identity
Every company has spent time to market their business. End-customers create an idea around the identity of the company they deal with, whether you plan for it or not.
voiceandvision makes the VUI fit in and effectively brands the CI within the new channel. - Team members & partners
During the development of a voice recognition-based application next to VUI designers, other specialists are responsible for infra-structure, programming and grammar design. With every speech-project there are speech platform (provider) and recognition issues to be cleared.
With a significant professional experience voiceandvision initiates and supports the organization and structure of voice-projects that lead to the right processes in order to make effective and efficient decisions.
Sample Dialogs & Wizard of Oz Test
Dialog strategies, prompt compositions, Sound & Feel all need to be tested with real users in a real situation. Checking and testing the initial concept of a possible dialog strategy in an early stage saves money, effort and time.voiceandvision carries out a professionally prepared and well-conducted WoZ test that provides valuable information in regard to what works, what might work or what doesn't work.
Design Strategy & Customized Recommendation
Defining the design strategy is the logical next step after requirements gathering and analysis. Interim results gathered need to be evaluated as they create the framework for the actual design phase. All the elements that contribute to the overall user experience need to be taken into consideration, such as event-handling and the use of non-speech audio.voiceandvision presents illustrative sample dialogs and explains possible dialog strategies and design decisions based upon the requirement gathering and analysis, real test data and profound evaluation.
Sound & Feel Definition
The quality of the VUI is partly achieved by a professionally designed Sound & Feel.“The idea is not to 'fool' callers by making them believe they are talking to a human being, but to make them forget they are talking with a machine” ©
Although the persona is a central component of the overall Sound & Feel, Sound & Feel is much more than just persona. It covers the entire impression callers get when interacting with the system like the VUI perspective, single- or multi-persona –approach, event-handling, gently handled error-recovery (psychological issues play a significant role), speech-style like prosody speech rate and phrasing that has its impact and affects the understandability of prompts, voice selection (casting) and use of non-speech audio.
voiceandvision creates a Sound & Feel definition that matches the design strategy and which is perfectly in tune with the data gathered during the requirement gathering and analysis phase.
Dialog Design
The dialog design covers the flow logic. It handles states of dialog and pre-conditions. In some cases an intelligent dialog design provides callers with a possibility to develop from novice- to power-users. Solid dialog design provides support on a need-to-know basis. The dialog designer carefully considers where to enable barge-in and when to allow the use of always-active commands.voiceandvision guarantees solid dialog design based upon the requirement gathering and analysis, real test data and profound evaluation and provides its customers with a specially adapted and detailed dialog design document that is easy to use by programmers and system-integrators.
Prompt Design
Prompt design is the core activity of VUI design. The composition of prompts needs to be extremely balanced. Information needs to be grouped in a logical way, arranged to meet the mind-set and mental models of callers. The dialog must initiate a conversational character in order to suit verbal communication modes. Talking with a machine does not automatically need to imply an unnatural form of communication. voiceandvision prompt design is thought in terms of dialog, how it takes place, analyzes human conversation and uses it.
voiceandvision knows and respects the difference between written and spoken language.
With professional expertise and profound experience voiceandvision composes well-considered prompts.
Voice Recording & Coaching
During the recording session the voice coach needs to make sure that the intention of the written prompt is captured correctly. The emphasis and phrasing within the spoken prompt must be perfect in order to optimize understandability and capture of the intended semantic meaning. Conversational aspects can only be captured during the recording session.
voiceandvision supports its customers with voice casting and coaches voice talents.
voiceandvision directs the recording sessions in order to guarantee the best possible transition from written text to recorded sound files.
Post Production
The result of recording sessions is raw material, where almost everything that took place during the session is recorded. Post-production starts with carefully selecting usable material that successively needs to be cut and batch-processed through several filters that optimize sound quality when prompts are finally sampled down from CD-quality (44.1 kHz) to the bandwidth of the telephone environment (8 kHz). The post-processing techniques depend on voice quality and pitch, as well as on service character and audio design. The recorded prompts represent the largest part of the entire Sound & Feel of the application and have a decisive role in regard to caller experience and acceptance, and with that the application's ROI.To guarantee professional standards, voiceandvision always controls the entire post-production phase.
Audio Design
Sounding logos (acoustic branding) have a significant impact as they address the hearers' fantasy and stimulate different processes within the human brain. Sometimes they even 'optimize' memorizing information. Non-spoken audio elements can provide and support a clear structure. In some cases a well-designed earcon informs more adequately than a spoken message. Audio design must always provide added value. Using audio elements just for the sake of it does not make any sense and could even be counter-productive.With professional expertise and profound experience voiceandvision carefully considers the use of non-speech audio. In case the design strategy initiates the use of non-speech audio, voiceandvision provides its customers with well-designed and produced environmental backgrounds, jingles, landmarks and earcons.
Testing
During the development of a voice recognition-based application, as in any other project, it's important to test early and often. Testing a VUI means both testing the implementation for correctness and testing usability and acceptance. After the requirement analysis a conceptual dialog strategy needs to be tested. This is done during a so-called Wizard of Oz test. The results of the WoZ-test indicate whether the intended dialog-strategy works well, it needs tuning, or that it needs to be re-evaluated.When the dialog design phase is completed the implementation can start. Functionality testing - with temporary prompts or TTS re-generated ones - supports the implementation. After recording and post-production the recorded prompts can be implemented. The system ‘becomes alive’ and with its final prompts again it needs to be tested in regard to acceptance and usability. Qualitative and quantitative test data is gathered by conducting standardized interviews and questionnaires. Log files are to be analyzed and field tests carried out, all in order to get a clear picture of the application’s status. Solid and frequent testing filters out irregularities during the development phase, supports its process and creates the possibilities for corrections in time and saves money.
With professional expertise and a specially adapted methodology voiceandvision initiates, supports and organizes testing of voice recognition-based applications.
Refinement Phase
During testing, details will come to the surfaces that need additional attention. Grammars need to be tuned in order to provide better recognition. Sometimes a wording does not work well or a pause is too long. Unforeseen back-end latencies need to be taken care of or need to be covered by correct prompting.voiceandvision supports its customers during the entire project and provides the additional VUI-related services required to bring out the best possible VUI.
Marketing & Rollout Support
A decision to automate with a speech recognition-based application requires a well-balanced and substantiated internal and external communication. Experience shows it isn’t always easy or obvious to point out the advantages such a system could bring.
voiceandvision communicates the essentials from business customer to end-user and back and is happy to share its solid project experience.
voiceandvision works closely together with marketing departments of its business customers in order to support efficient rollout strategy and optimized customer communication before launching the application into the market.