The aim of this course is to explain students the principles of voice technologies and to present practical applications both available currently and expected in the near future. The core of the lectures covers: methods concerning speech signal preprocessing, extraction of speech features, word and speech recognition, speaker recognition, as well as speech synthesis. The lectures are accompanied by exercises where students can work with real speech data and also analyze own voice.
Students attending the course gain basic theoretical knowledge about speech processing and will be able to implement simple systems and use them for practical purposes.
State-of-the-Art of speech processing, acoustic signals and their digitalization, speech signal, vowels, consonants, spectrum, spectrogram, speech signal features for speech recognition, selection of the best features, estimation of word endpoints, recognition of single words, statistic models for recognition of continuous speech, time alignment, speech signal features for speaker recognition, fundamental frequency of voice, text-dependent and text-independent speaker recognition, principles of speech synthesis, specific fields of speech analysis such as stress detection and alcohol detection.
Inhaltliche Schwerpunkte: Grundlagen digitaler Sprachsignalverarbeitung, Merkmale des Sprachsignals, Erkennung von einzelnen Wörtern, Erkennung kontinuierlicher Sprache, Sprechererkennung (Verifikation, Identifikation), Sprachsynthese (Text-to-Speech), Anwendungsfelder der Sprachverarbeitung, Experimentelle Untersuchungen.
Die erste Vorlesung des Teils 3 findet am Mittwoch, 1.3.2017, in Raum CG402 in der Zeit von 13.00 Uhr bis 16.00 statt. Weitere Vorlesungen des Teils 3 finden am 8.3, 15.3. und 22.3.2017 in Raum CG402 in der Zeit von 14.00 Uhr bis 17.00 Uhr statt sowie am 10.3., 17.3. und 24.3.2017 in Raum EI6 von 14.00 Uhr bis 17.00 Uhr (d.h. an insgesamt 7 Terminen im März 2017).