Communication between humans and machines is one of the biggest technology topics of our time. We take speech dialog systems like Siri or Alexa for granted in many areas of life today. The same goes for voice control systems in cars or translation tools like Google Translate. Speech technology can now analyze social media messages in real time for various purposes.
Computer scientist Wolfgang Wahlster from the German Research Center for Artificial Intelligence (DFKI) at Saarland University was already pondering the question of how computers can understand our human language in the 1990s. He and his team came up with the Verbmobil: a system that recognizes and analyzes spoken spontaneous language, translates it into a foreign language, then generates a sentence and says it aloud. The translation computer could be activated via cell phone by three-way conference. It spoke German, English, and Japanese and could interpret in seconds between businesspeople from Tokyo, New York, or Frankfurt who were speaking on the phone in their native language. To make this happen, Wolfgang Wahlster and his team had recorded human voice data for years and had developed implementation programs. They were assisted by around 100 computer scientists, language psychologists, linguists, translators, and other specialists who worked on the Verbmobil from 1993 to 2000. The main problems in its development were grammar and the ambiguity of many terms. The project received funding totaling 116 million deutschmarks from the Federal Ministry of Education and Research.
A computer that understands speech: the Verbmobil
The computer that understands speech set many other developments in motion, for example the largely voice-controlled car, computers that can read e-mails aloud, and automatic internet music searches using language terms. A Japanese company is even said to have used the technology to market a toy dog that searches for a ball and shakes paws on command.
In 2001, Wolfgang Wahlster was honored with the German Future Prize for his work on the Verbmobil.
(Image: Prostock-studio – AdobeStock.com)
