The aim of this thesis is to solve a stochastic inverse problem, related to a model for voice pro... more The aim of this thesis is to solve a stochastic inverse problem, related to a model for voice production, using artificial neural networks (ANNs). Th- ree parameters of the model are considered uncertain and random variables are associated to these parameters. For each random variable, a probabi- lity density function is constructed using the Maximum Entropy Principle. Substituting three uncertain parameters for the associated random varia- bles, the new model constructed is stochastic and its output is a stochastic process consisting of realizations of voice signals. The proposed inverse pro- blem consists of mapping the three random variables from the voice signals and the use of artificial neural networks to construct the solution of this in- verse problem. Features are then extracted from the output voice signals and taken as inputs of the designed ANN, which outputs are random variables. The probability density functions of these random outputs are estimated and compared with the original ones. Two kinds of problem are discussed. At first, the same probability distribution is used to generate the voice sig- nals and to solve the corresponding inverse stochastic problem. In this case, the actual probability density functions are very well fitted by the simula- ted one. Then, different probability distributions are used to generate the voice signals to be used to train the ANN and to solve the corresponding inverse problem. A pleasant surprise appears: the quality of the estimation is almost unchanged, except for one of the random variables. Keywords Inverse model. Stochastic process. Vocal folds. Neural Networks.
The aim of this thesis is to solve a stochastic inverse problem, related to a model for voice pro... more The aim of this thesis is to solve a stochastic inverse problem, related to a model for voice production, using artificial neural networks (ANNs). Th- ree parameters of the model are considered uncertain and random variables are associated to these parameters. For each random variable, a probabi- lity density function is constructed using the Maximum Entropy Principle. Substituting three uncertain parameters for the associated random varia- bles, the new model constructed is stochastic and its output is a stochastic process consisting of realizations of voice signals. The proposed inverse pro- blem consists of mapping the three random variables from the voice signals and the use of artificial neural networks to construct the solution of this in- verse problem. Features are then extracted from the output voice signals and taken as inputs of the designed ANN, which outputs are random variables. The probability density functions of these random outputs are estimated and compared with the original ones. Two kinds of problem are discussed. At first, the same probability distribution is used to generate the voice sig- nals and to solve the corresponding inverse stochastic problem. In this case, the actual probability density functions are very well fitted by the simula- ted one. Then, different probability distributions are used to generate the voice signals to be used to train the ANN and to solve the corresponding inverse problem. A pleasant surprise appears: the quality of the estimation is almost unchanged, except for one of the random variables. Keywords Inverse model. Stochastic process. Vocal folds. Neural Networks.
Uploads
Papers by Julien Mauprivez
Keywords
Inverse model. Stochastic process. Vocal folds. Neural Networks.
Keywords
Inverse model. Stochastic process. Vocal folds. Neural Networks.