Automatisk Identifiering av Inandningspauser i Spontant Tal - ett HMM/ANN-hybridsystem i Matlab
AndningInandningspauserNyckelordsidentifiering'HMM/ANN-hybridsystem'MatlabTaligenkänningProsodiTalproduktionAkustikFonetikLinguisticsAllmän språkvetenskap/LingvistikPhoneticsPhonologyFonetikFonologiTechnological sciencesTeknikLanguages and LiteraturesTechnology and Engineering
This thesis presents a system which has been implemented to satisfy a need in theresearch on how speech planning interacts with syntactic and prosodic structure inspontaneous speech. The long-term purpose of the research is to provide models forautomatic parsing of spontaneous speech and for psycholinguistical modelling of speechproduction. Identification of inhalation pauses is an important step in the developmentof automatic methods for spontaneous speech parsing.Identification of inhalation pauses is considered to be a keyword-spotting speechrecognition problem. Hybrid HMM(Hidden Markov Models)/ANN(Artificial NeuralNetworks) approach is applied to this problem. Method gets 90,8% in Recall, 66,4% inPrecision and 76,7% in F-score. Use of a threshold value for duration increases the Fscoreto 82,5%, therefore duration is considered to be relevant in performanceoptimization. Other proposed optimization parameters are better acoustic modelling,identification of the units causing false identifications prior to inhalation pausesidentification and production of a more appropriate spontaneous speech corpus.