In our current project, vocal signal has to be used to drive sound synthesis. In order to study t... more In our current project, vocal signal has to be used to drive sound synthesis. In order to study the mapping between voice and synthesis parameters, the inverse problem is first studied. A set of reference synthesizer sounds have been created and each sound has been imitated by a large number of people. Each reference synthesizer sound belongs to one of the six following morphological categories: " up " , " down " , " up/down " , " impulse " , " repetition " , " stable ". The goal of this paper is to study the automatic estimation of these morphological categories from the vocal imitations. We propose three approaches for this. A base-line system is first introduced. It uses standard audio descriptors as inputs for a continuous Hidden Markov Model (HMM) and provides an accuracy of 55.1%. To improve this, we propose a set of slope descriptors which, converted into symbols, are used as input for a discrete HMM. This system reaches 70.8% accuracy. The recognition performance has been further increased by developing specific compact audio descriptors that directly highlight the morphological aspects of sounds instead of relying on HMM. This system allows reaching the highest accuracy: 83.6%.
Questo lavoro presenta una tecnica per la stima del modello a due masse della corda vocale a part... more Questo lavoro presenta una tecnica per la stima del modello a due masse della corda vocale a partire da un dato flusso glottale tempo-variante. Il modello a due massee specificato da un certo numero di parametri meccanici di basso livello, calcolati in funzione di quattro parametri articolatori (livelli di attivazione di tre muscoli laringali e pressione subglottale). Le forme d'onda del flusso glottale, sintetizzate dal modello, sono caratterizzate da un insieme di parametri acustici per la quantificazione della sorgente vocale. ...
Abstract: This work presents a procedure for the estimation of a two-mass vocal fold model starti... more Abstract: This work presents a procedure for the estimation of a two-mass vocal fold model starting from a time-varying target ow signal. The model is specified by a large number of physical parameters, computed as functions of four articulatory parameters (three laryngeal muscle activations and subglottal pressure). Flow waveforms synthesized by the model are characterized by means of a set of typical voice source quantification acoustic parameters. Given a sequences of target acoustic parameters, dynamic programming techniques and ...
In our current project, vocal signal has to be used to drive sound synthesis. In order to study t... more In our current project, vocal signal has to be used to drive sound synthesis. In order to study the mapping between voice and synthesis parameters, the inverse problem is first studied. A set of reference synthesizer sounds have been created and each sound has been imitated by a large number of people. Each reference synthesizer sound belongs to one of the six following morphological categories: " up " , " down " , " up/down " , " impulse " , " repetition " , " stable ". The goal of this paper is to study the automatic estimation of these morphological categories from the vocal imitations. We propose three approaches for this. A base-line system is first introduced. It uses standard audio descriptors as inputs for a continuous Hidden Markov Model (HMM) and provides an accuracy of 55.1%. To improve this, we propose a set of slope descriptors which, converted into symbols, are used as input for a discrete HMM. This system reaches 70.8% accuracy. The recognition performance has been further increased by developing specific compact audio descriptors that directly highlight the morphological aspects of sounds instead of relying on HMM. This system allows reaching the highest accuracy: 83.6%.
Questo lavoro presenta una tecnica per la stima del modello a due masse della corda vocale a part... more Questo lavoro presenta una tecnica per la stima del modello a due masse della corda vocale a partire da un dato flusso glottale tempo-variante. Il modello a due massee specificato da un certo numero di parametri meccanici di basso livello, calcolati in funzione di quattro parametri articolatori (livelli di attivazione di tre muscoli laringali e pressione subglottale). Le forme d'onda del flusso glottale, sintetizzate dal modello, sono caratterizzate da un insieme di parametri acustici per la quantificazione della sorgente vocale. ...
Abstract: This work presents a procedure for the estimation of a two-mass vocal fold model starti... more Abstract: This work presents a procedure for the estimation of a two-mass vocal fold model starting from a time-varying target ow signal. The model is specified by a large number of physical parameters, computed as functions of four articulatory parameters (three laryngeal muscle activations and subglottal pressure). Flow waveforms synthesized by the model are characterized by means of a set of typical voice source quantification acoustic parameters. Given a sequences of target acoustic parameters, dynamic programming techniques and ...
Uploads