SpeechTEK Europe Keynoter Sees Changes Ahead
There are 13 ‘mega trends’ shaping the future of European speech, according to Professor Wolfgang Wahlster, who gave a keynote address at the first SpeechTEK Europe conference in London May 26-27. SpeechTEK Europe was launched to capitalize on the huge growth in speech innovation and implementation in Europe and, after the success of this first event, will run annually.
Wahlster, a member of the Nobel Prize committee and CEO of the German Research Center for AI, shared his vision of technology currently in development in laboratories and research institutes, technology that he predicts will be commercial realities in the next two to five years.
Trends he identified include the move from cooperative speech to spontaneous speech, the change from monolingual to multilingual systems, the development from database transactions to problem-solving dialogues, and the transition from accessing the ‘Web of information’ to ‘the Web of services.’
Among the many examples Wahlster spotlighted was a system for multiparty diarization and tracking, currently in development by AMI Consortium, which allows transcription and automatic summary of a teleconference, identifying who spoke when, with whom, and about what. This solution will be commercially available in two to five years. Wahlster described this as the ‘killer app for call centers’ since it can deal with overlapped, non-native, accented, and spontaneous speech and provides just-in-time access to relevant documents or fragments of past conversations.
Other examples he cited include age and gender recognition for personalized call center service. This can also be applied to automotive applications, such as tailoring the presentation to the age of the driver when he first enters the rental car.
Multimodal applications are becoming more and more important, according to Wahlster—and especially in the car. Research is now underway to look at how speech and eye gazes can be used together as input modalities to correct dictated speech messages while driving, for example.
According to Wahlster, challenges facing the speech research community in the next 20 years include integrating top-down contextual knowledge into low-level speech recognition processes, and exploiting the growing knowledge about human communication strategies derived from psychological and neurolinguistic research.
The opening keynote at SeechTEK Europe came from author and strategist Charlie Leadbeater, who spoke about the importance of mobilizing customers as co-creators and innovators. The Web means that consumers have access to a variety of social media tools that they can use to generate and share ideas, reviews, feedback, and product modifications. Dialogue with customers is vital for successful innovation, Leadbeater argued, and speech technology provides one method through which this dialogue can take place.
SpeechTEK Europe comprised two days of conference sessions and a parallel exhibition, with a day of practical, half-day pre-conference workshops covering the latest breed of speech technologies and their applications. The event welcomed hundreds of attendees from 25 countries, ranging from the Far East and Australasia, North and South America, as well as the whole of Europe. Some of the most popular conference sessions discussed using video in customer service; voice biometrics; multilingual applications; new uses of text-to-speech synthesis, and speech recognition techniques and experiences.
A sold-out exhibit hall ran in parallel to the conference, with sponsors including Voxeo, Autonomy, Avaya, Eckoh, Loquendo, NICE, Openstream, and Verint.
Also at the event, the winners of SpeechTEK Europe’s Avatar Challenge were announced. Finalists in this brand new competition comprised seven avatars developed for use in a range of business applications, speaking a variety of languages, and using different text-to-speech engines, animation and modeling techniques, from fantasy cartoons to realistic human-like appearance.
Voting took place in two separate polls - The Experts’ Prize and The People’s Choice Award. Experts’ Prizes went to ejTalk and Humanity Interactive, and the People’s Choice Award went to H-Care. In addition, a Judges’ Special Award was given to VoxWeb
The entries were judged by user interface experts, assisted by foreign language experts, using the following criteria: appearance and movement, voice expression, use of facial and body gestures, and voice, lip and facial synchronization.