Adaptive estimation of speech parameters

Date
1994-03
Journal Title
Journal ISSN
Volume Title
Publisher
Stellenbosch : Stellenbosch University
Abstract
ENGLISH ABSTRACT: Linear predictive coding(LPC), and transformations of it, is currently the most popular way of analysing speech signals. Major limitations of using a frame-based technique are that each frame is analysed in isolation of the rest while assuming the excitation source to be a white, gaussian process. In order to reduce computation time, an all pole model is usually employed. In this project an adaptive algorithm is proposed for speech signal analysis. The algorithm is based on the recursive least mean squares method with a variable forgetting factor. A pole-zero model is used to: estimate the anti-formants present in certain sounds (i.e. nasals and nasalized vowels). This method offers better detection of poles and zeros in stationary environments and faster tracking of pole and zero frequencies in nonstationary signals than other sequential methods. An effective input estimation algorithm eliminates the influence of pitch on the parameter estimates by assuming the input to be a white noise process or a pulse sequence.
AFRIKAANSE OPSOMMING: Linieere voorspellings-kodering, en transformasies daarvan, is huidiglik die gewildste metode t.o.v. die analise van spraakseine. Blok-gebaseerde algoritmes het ernstige tekortkominge. Elke raam word byvoorbeeld in isolasie van die res geanaliseer terwyl daar aangeneem word dat die intree na die spraakkanaal 'n wit, gaussiese ruisproses is. Om berekeningstyd te beperk word 'n model met slegs pole gebruik. In hierdie projek word 'n aanpasbare algoritme (gebaseer op die rekursiewe kleinste kwadrate metode) met 'n varierende vergeetfaktor voorgestel. 'n Pool-zero model bied akkurater opsporing van pole en zeros in stasionere omgewings. Dit bied ook vinniger volging van pool en zero frekwensies in nie-stasionere seine as ander aanpasbare algoritmes. 'n Effektiewe intree-beramings algoritme skakel die invloed van die fundamentele frekwensie op die beraamde parameters uit. Dit word reggekry deur te aanvaar dat die intree 'n wit ruis-proses of 'n pols reeks kan wees.
Description
Thesis (MEng)--University of Stellenbosch, 1994.
Keywords
Speech processing systems, Automatic speech recognition, Algorithms
Citation