ALADIN paper @ to be presented at NordiCHI
We will be presenting a full paper on the first phases of the ALADIN design process at the NordiCHI conference, 26-30/10/2014, in Helsinki. The paper is about how we approached the user-centered design of the self-learning ALADIN speech engine, trying to find out how people address such a speech system, and which words people use to talk to it.
Just to give you an idea, you can find the (preliminary) abstract below:
This paper describes the user-centered design of ALADIN, a speech recognition system targeted at people with physical disabilities, many of whom also have speech impairments. ALADIN is a self-learning system, designed to allow users to use their own specific words and sentences, adapting itself to the specific voice characteristics of the user. The design process described in this paper focuses on the specific interaction issues associated with this type of voice interaction. Specifically, the tests focused on how users with speech impairments address a speech interface, determining which types of variation in wording and sentence structure occur, and when they occur. The results provide a detailed analysis of the observed variation. Based on these results, we discuss potential causes of this variation, and how the users’ expectations can better match the capabilities of the ALADIN speech system through careful interaction design.
ALADIN Demo @ Speech Technology and Human Computer Interaction Workshop
On Thursday, March 27th, we demoed the ALADIN demo to the participants of the Speech Technology and Human Computer Interaction Workshop hosted at Edinburgh University. Besides demoing the ALADIN system, attending interesting presentations and workshops, we also managed to get a few workshop participants to train ALADIN. Since the workshop, we know ALADIN has no problem with speech input in Scottish and (Mandarin) Chinese!
Video: Aladin at iMinds the Conference ’13
On december 5, 2013, Mobilab and CUO | Social Spaces presented a demo of the ALADIN project at iMinds the Conference, showing off the vocal interface for domotics in both standard Dutch and West-Flemish. iMinds The Conference is a yearly conference bringing together the best and brightest in academia, policy making, business and venture capital in Belgium, with keynote speeches from thought leaders and breakout sessions on collaborative research and entrepreneurship. This year’s theme was “digital innovation in Europe”, and besides presentations there were over 40 demos from fields such as smart cities, media, future internet, or health.
Our demo was the first public demonstration of the ALADIN vocal interface engine in action, making use of the 3D environment used in our user tests, a playstation controller for movement and as a start/stop button for speech commands, the companion iPad application and a home automation box with three connected lamps.
Because the online learning is not yet operation-ready, we trained the system beforehand, repeating each command 6 times (3 times in a noisy environment, 3 times in a quiet environment) using a close talk microphone, which was also used during the conference. During the set-up and plenary meetings at the conference, when things were relatively quiet, we achieved an almost perfect accuracy in recognition, but more surprisingly, the system managed to maintain a high accuracy (80%) during the noisy break periods. Only during the live performance of a pumkin orchestra on the demo floor did the accuracy drop below 50%.
If you missed the conference, you can find a video of our demo on top.
Demos of AToM and ALADIN at iMinds The Conference
At the annual conference of iMinds, iMinds The Conference, the results of two of our health-related projects will be demonstrated.
The demo of the AToM project consists of a smartphone app and a tablet app for dementia caregivers. These apps give caregivers access to dynamic meal information (e.g. diets, medication, need for assistance, but also personal preferences and recommendations) and to evaluate the meal experience, generating feedback for future meals.
The ALADIN demo is a 3D, interactive home environment that is operated with an assistive voice control system. It shows how people with physical disabilities can use speech to open doors and window blinds, or rely on a customized touch screen interface that offers maximal efficiency and ease-of-use.
In total, more than 40 demos will be showcased and the conference program includes several interesting international speakers. Please check out the full program at the conference’s website, where you can also register to attend.