Modeling of Phone Features for Phoneme Perception

Konferenz: Speech Communication - 12. ITG-Fachtagung Sprachkommunikation
05.10.2016 - 07.10.2016 in Paderborn, Deutschland

Tagungsband: ITG-Fb. 267: Speech Communication

Seiten: 5Sprache: EnglischTyp: PDF

Persönliche VDE-Mitglieder erhalten auf diesen Artikel 10% Rabatt

Hoege, Harald (Universität der Bundeswehr München, 85577 Neubiberg, Germany)

Neuronal feature extraction is performed along the auditory pathway. The features are generated from spectro-temporal analysis of modulation spectra of the auditory signal provided by bundles of hair-cells. We postulate, that in the auditory cortex the spectro-temporal features are transformed to categorical features called phone features. Whereas the nature of the spectro-temporal features is well explored the nature of the categorical phone features is widely unknown. There are indications, that the categories are related to place and manner of articulation. We develop a statistical model for the phone features, which is consistent with psycho-acoustic findings and measurements of neuronal activities. We use this model to simulate phoneme perception. Although the current model shows a large gap to human performance, the results achieved allow to draw conclusions concerning further directions in research.