T´tol ı hmm-based speech synthesis applied to spanish and english, its applications and a hybrid approach

C.I.F G: 59069740 Universitat Ramon Lull Fundaci´ Privada. Rgtre. Fund. Generalitat de Catalunya n´ m. 472 (28-02-90) . o u

TESI DOCTORAL
T´tol ı HMM-based speech synthesis applied to Spanish and English, its applications and a hybrid approach Xavier Gonzalvo Fructuoso Escola T` cnica Superior d’Enginyeria Electr` nica i e o Inform` tica La Salle a Comunicacions i Teoria del Senyal Dr. JoanClaudi Socor´ Carri´ o e Dr. Ignasi Iriondo Sanz

Realitzada per en el Centre

i en el Departament Dirigida per

C. Claravall, 1-3 08022 Barcelona Tel. 936 022 200 Fax 936 022 249 E-mail: urlsc@sec.url.es www.url.es

Abstract

Nowadays, Human Computer Interface (HCI) is one of the most studied disciplines in order to improve real human interactions with machines on the present time andfor the incoming future. More and more electronic devices of the daily life are used by more people. This electronic incursion is mainly due to two reasons. On the one hand, the undoubted increasing of the economical accessibility to this technology but on the other hand, the more friendly interfaces that allow an easier and more intuitive use. As a matter of fact, nowadays it is only necessary toobserve the personal computer interfaces, pocket size computers and even mobile telephones. All these new interfaces let little experienced users make use of cutting edge technologies. Moreover, the inclusion of speech technologies in these systems is becoming more usual since speech recognition and synthesis systems have improved their performance and reliability. The purpose of speechtechnology is to provide systems with a natural human interface so the use can be extended to daily life. Text-To-Speech (TTS) systems are one of the main modules under intense research activity in order to improve their naturalness and expressiveness. The use of synthesizers has been extended during the last times due to the high-quality reached in real limited domain applications and the good performancein generic purposes applications. However, there is still a long way to go with respect to quality and open domain systems. This work will present a TTS system based on a statistical framework using Hidden Markov Models (HMMs) that will deal with the main topics under study in recent years such as voice style adaptation, trainable TTS systems and low print databases. Moreover, a cutting edgehybrid approach combining concatenative and statistical synthesis will also be presented. Ideas and results in this work show a step forward in the HMM-based TTS system field.

Resumen

Hoy en d´ la Interacci´n Hombre-M´quina (IHM) es una de las disciplinas m´s estudiadas con el ıa, o a a objetivo de mejorar las interacciones humanas con sistemas reales para el presente y para el futurovenidero. M´s y m´s dispositivos electr´nicos son usados por m´s gente en la vida diaria. Esta a a o a incursi´n electr´nica se debe principalmente a dos razones. Por un lado, el indudable aumento en o o la accesibilidad econ´mica a esta tecnolog´ pero por otra parte, unos interfaces m´s amigables que o ıa a permiten un uso m´s f´cil e intuitivo. Simplemente hace falta observar hoy en d´ los ordenadores aa ıa personales, las computadoras de bolsillo e incluso los tel´fonos m´viles. Todos estos nuevos disposie o tivos admiten que usuarios poco experimentados puedan hacer uso de las tecnolog´ m´s punteras. ıas a Por otra parte, la inclusi´n de las tecnolog´ del habla est´ llegando a ser m´s com´n gracias a o ıas a a u que los sistemas de reconocimiento y de s´ ıntesis de voz han estado mejorando sufuncionamiento y fiabilidad. El objetivo final de las tecnolog´ del habla es crear sistemas tan naturales como los seres ıas humanos para que su uso se pueda extender a cualquier rinc´n de la vida diaria. Los conversores de o Texto-a-Voz (o sintetizadores) son de los m´dulos que m´s esfuerzo investigador han recibido con el o a objetivo de mejorar su naturalidad y la expresividad. El uso de los...