Please use this identifier to cite or link to this item: https://www.um.edu.mt/library/oar/handle/123456789/92283
Title: Intonation analysis and modelling of speech
Authors: Borg Ebejer, Alison (2010)
Keywords: Speech processing systems
Natural language processing (Computer science)
Automatic speech recognition
Issue Date: 2010
Citation: Borg Ebejer, A. (2010). Intonation analysis and modelling of speech (Bachelor's dissertation).
Abstract: As the quality of speech synthesis improves, a greater demand is put on the intonation system to produce more varied intonation tunes. Every society has its own intonation. In Malta our intonation characterizes the way we Maltese elucidate the Maltese language. There are many intonation models; however an intonation model of the Maltese language is still being developed. This thesis is the development of an intonation model on stressed syllables for the Maltese language. A five minutes speech is used as a basis for the development of the model. The project required studying the pitch patterns of the stressed syllables. These were then reduced to a set of approximations. These pitch patterns were used for the development of the intonation model. The two main software packages that were used throughout this thesis were PRAAT and MATLAB. PRAAT was used for the annotation of speech and for bringing out the pitch listings of every stressed syllable that was used for analysis, whilst MATLAB was used to write the program for the intonation model. The results obtained were quite satisfactory. This is evident by the mean square error that was calculated between the PRAAT and the intonation model. 43.75% all the stressed syllables found in the testing phrase gave a 'very good match'. 3 7 .5% of all the stressed syllables found in the testing phrase resulted in a 'good match'. The consistency of the model was also calculated. This resulted in 50% 'very good consistency' and 50% 'good consistency'. There was no score in the 'poor consistency' category. Future work can be done on more resources (speech) in order to achieve a more detailed analysis. Apart from this, the analysis can also be made on the unstressed syllables and thus the intonation model will cater for both stressed and unstressed syllables.
Description: B.SC.(HONS)COMPUTER ENG.
URI: https://www.um.edu.mt/library/oar/handle/123456789/92283
Appears in Collections:Dissertations - FacICT - 2010
Dissertations - FacICTCCE - 1999-2013

Files in This Item:
File Description SizeFormat 
BSC(HONS)ICT_Borg_Ebejer_Alison_2010.pdf
  Restricted Access
12.98 MBAdobe PDFView/Open Request a copy


Items in OAR@UM are protected by copyright, with all rights reserved, unless otherwise indicated.