Agora: "VoiceÄpp" identifies dialects thanks to speech recognition technology
"Gigertschi", "Gröibschi", "Bütschgi" or "Murmutz"? Swiss German dialects show great variation - linguists at the universities of Zurich and Geneva have made an app that identifies dialects through speech recognition. The project was funded by Agora, the SNSF scheme for science communication.
On Monday, 5 January 2015 around thirty people met at the Department of Comparative Linguistics of the University of Zurich to celebrate the launch of "VoiceÄpp". The application, which is the result of an SNSF Agora project, can swiftly identify a dialect as well as the speed of speaking and the voice pitch thanks to speech recognition technology.
"Abig", "Oobig" or "Aabe"
To identify a dialect, users say 15 significant words into the microphone of a smart phone; the app then determines where the dialect of the speaker originates from - more or less down to the exact village. This function also makes it possible to test how well you can imitate a dialect. The voice analysis feature is able to illustrate how people with hearing difficulties, e.g. tinnitus, hear their own voice. In addition to dialect and voice analysis, the app provides ample information about dialects and phenomena related to speech recognition. The "VoiceÄpp" also offers a dialect panorama feature - like the app that is able to identify mountain tops in a panoramic view: if users hold up their smart phone, the app will highlight the pronunciation of certain words in the vicinity.
Adrian Leemann, project leader of "VoiceÄpp" and former postdoc at the Phonetics Laboratory of the University of Zurich, hopes that users of the interactive app will playfully gain an understanding of linguistic research and what it is all about. As everyone speaks a dialect, the application should have a large audience among Swiss German speakers and it holds the potential to introduce linguistic research to people who are unlikely to be aware of this topic otherwise. Two weeks after its launch, the app has already been downloaded more than 15´000 times.