More information on cookies
Over 2 years, Marc Mersiol has been leading the Orange Vocal TV interface design. The lessons learned from this innovation project has offered him a broad learning experience and widen his vision of design practice.
This project covers half of our Orange TV clients that allows voice enabled interactions with their TV. By simply pressing the microphone button on their remote control and formulate their request in a natural language.
The system will translate the request and provide a prediction-based response. It can be for example to switch to a live channel, to initiate a program search, to display the evening TV program, to launch any service like a VOD catalogue, to change the sound level, or even to control the connected home accessories.
Our team has worked around the design experience, the users’ journey, and the conversation trees.
Starting from a shared defined experienced, the first step was to agree on the minimum viable key functionalities for the users and stay on them. This allowed us to easily arrive to a technical solution that was possible to quickly deploy with our customers.
Once the customer usage began, we were able to collect and analyze the user’s data. In a continuous enhancing process, everything was optimized: the processing models, the speech understanding and the response systems to the user’s command.
This experience was enriching due to a new methodological adoption approach, Notwithstanding the objective remains the design and building the best user experience, the roads to achievement were certainly different.
Before, we used to start from an ideal experience definition and the desirable user’s journeys, where the graphic interfaces were the main goal. Today, in the vocal fields, the graphic interfaces lose their relevance. All work is concentrated on a dialogue-based system design, in which, through the user’s commands, the system will deliver the answers creating the customer’s satisfaction. To achieve the best dialogue design, it is a must to start out from the dialogue usage analysis where understanding of the user’s needs will be on the center stage.
The designer invents the human-machine interaction for machines to meet the user’s needs. For this, the designer will consider the users’ requirements while the artificial intelligence translate the correct responses. It is still a Ux design work, with the designer’s aim to invent intangible vocal interfaces rather than of foreseeing graphic interfaces.
At the beginning of the project, developers asked: “give us new screens”, where instead my activity consisted in adding a vocal layer on top of the existing screens! We need to build a new way of working together with the technical teams.
Working in vocal field implies a paradigm swift: Technology serves the users and remains mainly invisible.
Users knowledge is mandatory in vocal design: the UX designer should examine usages to build the analytical models and speech understanding instead of reinventing graphic interfaces. This will enable AI to deliver the expected response to the users.
There is a massive amount of hidden work requiring daily analysis of the user’s request. The model allows the system to acknowledge the users’questions thus continuously being shaped with both the “data scientists” and “data analysts”. The objective is to adapt the system response in a way to attain a success rate near to 100%.
Never start with a hunch nor from technical solutions point. To work on a vocal interface means to start from a blank sheet with no limits or constraints of classical interactions, like graphical interfaces. The system delivered is much more open, elastic, and flexible. Sometimes, it might look impossible.
Also, rather than enforcing functionalities that users dislike, it is vital to start from their real needs and usages. The agile methodologies now become obvious, you should start as fast as possible with a minimal solution but functional, a Minimal Viable Product (MVP) to place in the user’s hands. And very quickly collect the users’ data, analyze it, and enhance treatment models and speech analysis. It is a fastidious and repetitive task, but this is the cost for a system to work with a “magical” touch for customers and ensure its success.
These interfaces are mainly complementary and not opponents. Voice allows spontaneity and immediateness. But, for example, if we offer over 7 different TV channels programs by vocal answers to an individual at the same time, the memory is unable to retain them. In this case, a visual display is the most appropriate. The work of the designer is partly to select between the different ways of communicating. To help construct a Human-Machine dialogue by vocal questions, we have realized a pictogram display in the screen to confirm the system is listening, while a spinning wheel will indicate that the user’s request is being reviewed.
The pictogram display in the screen confirm the system is listening
The spinning wheel indicates that the user’s request is being reviewed
The vocal design opens up all kinds of possibilities for the digital experience designer. The designer is no longer eyeballed on static graphical interfaces from the start as completed, Instead, he can invent and imagine infinite experiences that evolves over time and in function to the users’ desires.
In return, the UX designer will have to accept to work indifferently, spending much more time analyzing the data. It is a very new approach for the designer but essential, as soon, the vocal interactions in conjunction with the artificial intelligence, will be without doubt, part of our everyday lifestyle.
Discover below the videos on interaction design:
What is the design of interaction?
Key dates in interaction design
Testimonies on interaction design and multimodality
we want to know your view about design.orange.com