The beginning of the activities in speech processing at the ETH dates back to 1976. The first projects were about speech coding and noise cancellation. In the eighties, the scope was successively changed to new topics, namely text-to-speech synthesis, speech recognition, and speaker verification (in German: Sprachsynthese, Spracherkennung und Sprecherverifikation).
Towards the end of the nineties, the research in speech synthesis, that concentrated to German so far, was extended to other languages and particularly to polyglot TTS synthesis, i.e., the processing of mixed-lingual texts, e.g., German texts with various types of inclusions from other languages, mainly from English, French and Italian.
Additionally, there were increasing activities in forensic speaker verification case work, mainly since 1996.
For specific information about the work of the speech processing group, please go to the projects site.
The group offers a two-semester course in speech processing which consists of two hours of lecture and two hours of laboratory work (computer-base exercises). For information about the content of this course, please consult the text book and the accompanying material (in German only).