Workshop on Making Speech Mainstream: Problems and 10 Recommendations
Introduction: The Workshop for Speech Executives
At SpeechTEK 2005, speech industry executives gathered in a three hour workshop to discuss how to bring speech technology into the mainstream. The workshop attendees included executive representatives from speech users, service providers, platform providers, software providers and tool providers. Intel's Tim Moynihan organized and moderated the session.
The discussion was lively and spirited—it was clear that those who participated felt strongly that speech can solve real business problems and were passionate about making this technology more broadly deployed. There was immediate agreement that speech has not yet reached the mainstream. The following summarizes the problems identified and presents recommendations made by workshop attendees.
Problem: Speech needs to be easier to develop and deploy. Development is complex—professional services are required, for even simple applications. There is a lack of skilled speech professionals—there are not yet enough certified or talented speech application development professionals. Deployment costs are high—while platform license prices have declined slightly, the increased complexity of speech applications results in increased professional service hours resulting in an overall increase of application cost.
Recommendation 1: Speech technology trainers and educators should provide the best possible training to decrease the effort and improve the quality of speech applications for all phases of speech application development—application justification, design, implementation, usability testing, deployment, and monitoring. Industry associations should create and promote a curriculum for training speech application developers worldwide with the skills to design, prototype, test, and deploy speech applications. Educators should highlight a broader range of success stories and educate speech application developers on successful development and implementation methodologies across a broader range of industries.
Recommendation 2: Streamline the process for designing, implementing, usability-testing, deploying, and monitoring speech applications. Application developers should develop and share methodologies and best practices for developing speech applications. User interface experts should define clear and consistent usability metrics for measuring the quality of speech applications. Tool vendors should identify and incorporate best practices in speech application development tools and reusable code. Standards bodies should specify interfaces enabling development tools to exchange data and work together.
Recommendation 3: Decrease the startup costs of speech applications. All of the players in the value chain need to make speech more affordable. Speech technology vendors should lower their license fees; platform vendors should discount the cost of speech platforms; the professional services should make better use of development tools and reusable code to decrease their application development fees; and usability testers should streamline the testing process to decrease the cost of usability testing. Decreasing the cost of speech applications will increase the market size and result in a net increase of profits for all.
Problem: Speech deployments are focused on the contact center and are only slowly propagating throughout enterprise IT. Speech has been primarily sold into enterprise contact centers—there is a lack of understanding of speech benefits. Speech is not recommended for wide area workforce—the value of speech has not yet been clearly demonstrated as a key element in IT's strategy to enable a remote work force. Speech cannot easily connect to mainstream IT architecture—while standards have reduced the complexity of integrating speech platforms into the mainstream, it can still be challenging. Speech models have not yet adapted to Web models—speech applications need to integrate better to Web service frameworks and must get a larger mindshare from IT. Speech must occupy a larger mindshare of IT professionals focused on Web applications—many IT professionals concentrate on GUI interfaces and do not realize the advantage of VUI, especially for the mobile workforce. Lack of customer automation strategy—many enterprises still don't have a clear strategy; nor are there services available to develop an integrated Web/speech strategy. Lack of new, innovative speech applications—develop new applications that generate revenue.
Recommendation 4: Industry associations, service providers, platform providers, software providers and tool providers should promote speech throughout enterprise IT. Industry associations should publicize success stories, case studies, and provide speakers, articles and marketing collateral to promote speech through enterprise IT.
Recommendation 5: Platform and service providers recognize and support customer diversity by enabling alternative implementation models that fit the diverse needs of customers. Explain the advantages and disadvantages of premise-based, hosted/outsourced or a hybrid service model and assist customers in choosing the right model for their needs. This could be done through case studies and through education at industry events.
Problem: Advantages of speech are not clear. Speech needs a clear value proposition—the industry often uses only ROI as the key metric.
Recommendation 6: Industry analysts should develop criteria and metrics for measuring the success of speech applications. By systematically applying criteria and carefully evaluating the resulting metrics, speech applications will improve in quality, consistency and accuracy of information dissemination, increased brand loyalty, improved customer service, and improved employee productivity.
Recommendation 7: Industry analysts should develop business models for revenue generation. Business models (automation vs. cross-selling) need to be re-thought—some industries do not want to provide a rich self-service environment, since they see human interaction as a way to cross- and up-sell customers. Industry associations should develop and publicize models for initial product selling, cross-selling, upselling, and improving customer care using speech technologies.
Recommendation 8: Enterprises should leverage the changing environment. Industry analysts should explain how VoIP will increase the demand for speech and how this will impact organizational structure and procedures. Industry analysts should provide clearer definitions on how speech can play a role in mobilizing workers to increase productivity and contain costs. Multimodal application vendors should provide better value propositions for multimodal applications. Business planners should incorporate larger technology trends highlighting importance to our planning process as an industry.
Recommendation 9: Conference and event organizers create a refreshed market place venue. The organizers of speech conferences and trade shows should focus as much on customers as on interaction among members of the speech community. They should invite consumer and brand product managers and have integration panel discussions to highlight success and articulate concerns to industry executives on speech technology. The organizers of speech conferences and trade shows should expand to include a broader group of vendors. They should broaden the attendees and design events which give incentive for more customers to attend. Perhaps it makes sense to target some speech events more at vendors, and others more at customers/buyers of the services/products.
Problem: Poor speech experiences. There are not enough cool, effective speech applications - where's the cool factor? Speech vendors should create a virtual tour of world-class speech applications. Speech associations should post innovative student projects to their voice portals to inspire developers to adopt these innovations while being mindful of functionality requirements.
Recommendation 10: The speech industry should form a collaborative body similar to an industry association. The speech industry should formalize efforts through a consortium to describe, discuss and recommend industry actions. The focus should be on application successes, education and benchmarking to provide better understanding of the value of speech and linkage to IT concepts for decision makers.
We plan to meet again at SpeechTEK West in February to check the progress of these recommendations and to formulate specific plans to implement specific recommendations.