Automatic Speech Recognition Seminar
Automatic Speech Recognition Seminar
A seminar in collaboration with the Signal Processing group.
Contents:
In this seminar, we address all aspects of automatic speech recognition, i.e. the recognition and transcription of spoken language. This starts on the level of the audio signal, where we extract features and process signals in order to achieve a high robustness with respect to noise and speaker variation. Apart from robust approaches to training we will also cover the recognition of phonemes and words, up to automatic correction on the basis of natural language semantics.
Students present selected topics in order to achieve an overview on relevant techniques in the area of speech recognition.
Goals:
In this seminar, students learn
- reception and critical acclaim of research papers
- presentation techniques
- theory of automatic speech recognition
- aspects of machine learning
- aspects of signal processing
- aspects of language technology
Procedure:
After a general introductiion to goals and the structure of the seminar, we will distribute topics. Every participant will hold a presentation on one topic in automatic speech recognition. We expect active participation in the subsequent discussions. At the end of the seminar period, we expect a short summary paper.
Literature:
individually assigned per topic
More information on grading:
To successfully pass, we ask the following:
- seminar presentation on one topic in automatic speech recognition
- summary paper
- active participation in the discussions