From smartphones and automatic translators to domotic systems, over the past few years, some subfields of Artificial Intelligence as the natural language processing and speech synthesis, have become so much widespread that they are more and more present in our everyday life. Their essential aim is to simplify the communication between man and machine but, in order for a computer to communicate with a human being, it should be able to understand natural language (human language). The following thesis intends to provide an overview of techniques underlying natural language processing and speech synthesis. After presenting a general introduction about Artificial Intelligence and algorithms of machine learning and deep learning, they were described the various phases of natural language processing designed to eliminate the several ambiguities that characterize human language. Subsequently, the third chapter dealt with the generation of natural language aimed at the realization of a textual discourse and finally, the last point was the vocal synthesis that provides a spoken output of the generated text, and more, through the artificial recreation of the human voice.
Negli ultimi anni, l'elaborazione del linguaggio naturale e la sintesi vocale, sottocampi dell'intelligenza artificiale, si sono diffuse notevolmente tanto da essere sempre più presenti nell'utilizzo quotidiano, dagli smartphone e traduttori automatici fino ai sistemi di domotica. Il loro obiettivo essenziale è quello di agevolare la comunicazione tra uomo e macchina ma, affinché un computer possa comunicare con l'essere umano, deve essere in grado di comprendere il linguaggio naturale (umano). La seguente tesi si prefigge di offrire una panoramica sulle tecniche alla base dell'elaborazione del linguaggio naturale e della sintesi vocale. Dopo aver presentato un'introduzione generale al mondo dell'intelligenza artificiale e agli algoritmi di machine learning e deep learning, sono state descritte le varie fasi dell'elaborazione del linguaggio naturale atte ad eliminare le numerose ambiguità che caratterizzano il linguaggio umano. Successivamente è stata trattata la generazione del linguaggio naturale finalizzato alla realizzazione di un discorso testuale fino ad arrivare al passaggio successivo, la sintesi vocale che fornisce un output parlato del testo generato (ma non solo) attraverso la riproduzione artificiale della voce umana.
ELABORAZIONE DEL LINGUAGGIO NATURALE E SINTESI VOCALE
GADDI, STEFANIA
2020/2021
Abstract
From smartphones and automatic translators to domotic systems, over the past few years, some subfields of Artificial Intelligence as the natural language processing and speech synthesis, have become so much widespread that they are more and more present in our everyday life. Their essential aim is to simplify the communication between man and machine but, in order for a computer to communicate with a human being, it should be able to understand natural language (human language). The following thesis intends to provide an overview of techniques underlying natural language processing and speech synthesis. After presenting a general introduction about Artificial Intelligence and algorithms of machine learning and deep learning, they were described the various phases of natural language processing designed to eliminate the several ambiguities that characterize human language. Subsequently, the third chapter dealt with the generation of natural language aimed at the realization of a textual discourse and finally, the last point was the vocal synthesis that provides a spoken output of the generated text, and more, through the artificial recreation of the human voice.I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.14253/85988