Conversational Interfaces are truly becoming a standard in our everyday lives.
As we become more comfortable with this way of interacting with tech, our expectations increase. As users we now increasingly expect the companies we interact with to be available 24-7,
not only through a chat window on a website – but through the channel and device that suits us at a particular moment.In the car, on my mobile, sometimes through text, sometimes Voice (smart speaker or phone etc).
Earlier this year we embarked on a project for Södertälje municipality, delivered in cooperation with Softronic and Boost.ai. The purpose was to (a) explore the quality of speech-to-text when interacting with a digital assistant through phone, (b) understand the specific aspects of Voice Design compared to designing for text interaction and (c) testing its implementation by implementing Voice Design on an existing chatbot solution – ‘Kommun-Kim’ – thus enabling reusing earlier investment into its text-based solution and managing the solution with one ‘source of truth’.
“Conversational AI voice applications is a game changer. Imagine how we can increase accessibility to people that have difficulties reading and writing. Accessibility is one of Södertäljes key strategic overarching development areas”, says Tony McCarrick, Digitalization Strategist at Södertälje municipality.
As for the quality of speech-to-text, in this case done in Swedish and utilizing a solution developed by the German company Lindenbaum, the results of this tests showed very satisfactory results though a large-scale test was not part of the scope at this time. Some aspects, such as understanding dialects, can still pose a challenge but is in general not a significant issue with modern technology.
As for specific aspects of Voice Design, Talking to me utilized its proprietary TALKS ™ process to identify and implement Voice Design Adaptions in order to, for example, handle responses involving visual elements and adjustments involving securing the conversation flow through Voice (for instance; ‘end all bot utterances with a question’). The other aspect involved conducting Voice Content Adaptions involving aspects such as dividing content into smaller chunks and implementing specific intents needed to enhance the Voice experience.
As for the implementation – the Voicification was initially conducted over a limited series of intents relating to a specific topic within Södertälje municipality’s general information.
”I am very pleased with the process, cooperation and results of this project, not the least in terms of the team’s ability to deliver quality on time. We clearly showed that the technology for AI-driven conversations through Voice is mature and can be made accessible through phone and smart speakers. We also learned a lot around the importance of utilizing specific design competence to deliver the best user experience.”, says Tony McCarrick.
Listen to the a sample of the interaction here (in Swedish).
