Pɛtɾʊs is a an online automatic phonetic transcription system for Brazilian Portuguese. For example, Pɛtɾʊs automatically converts a sequence of letters like "descrédito" into a sequence of phones [ʤiskɾɛʤɪtʊ].
In order to represent the pronunciation of a word, the IPA (International Phonetic Alphabet), a latin-based alphabetical system for phonetic notation, is utilized. Beyond phonetic transcription, Pɛtɾʊs offers more information to users like: stressed syllable identification and syllabification. Also, it transcribes hetereophonic homographs (words with the same spelling, but different pronunciation), generating both forms as output.
If you would like to transcribe a single word or a file, just go to the main website.
The syllabic division consists in the identification and delimitation of the syllables that constitute each word. To perform this task automatically by Pɛtɾʊs, we adopted two algorithms: a rule-based algorithm described in Silva (2011) and the CECI algorithm.
The syllabification algorithm proposed by Silva (2011, p. 39) was developed on the basis of the Brazilian Portuguese spelling, but predominating phonology of words. Thus, the words "obstrução" (occlusion), "arredondar" (to round) and "assado" (roasted) are divided into syllables as follows, respectively: (o.bs.tru.ção), (a.rre.don.dar) and (a.ssa.do).
The automatic phonetic transcription called Pɛtɾʊs is a project carried out in "Instituto de Biociências, Letras e Ciências Exatas da UNESP" (IBILCE), by the doctoral student in linguistic Vanessa Marquiafavel Serrani, under the supervision of Prof. Dra. Claudia Zavaglia. Pɛtɾʊs is dedicated to the automatic phonetic transcription of simple words from Brazilian Portuguese, having as reference São Paulo state standard pronunciation.
- Test by word using the Silva (2011) algorithm:
$ python test_word.py -s silva -w guerra
guerra -> [ˈge.xa] | gue-rra | [gue]-rra
- Test by word using the CECI algorithm:
$ python test_word.py -s ceci -w guerra
guerra -> [ˈge.xa] | guer-ra | [guer]-ra
- Example transcribing transcribes Hetereophonic Homographs (HHs):
$ python test_word.py -s silva -w molho
molho -> [ˈmo.ʎʊ, ˈmɔ.ʎʊ] | mo-lho | [mo]-lho
- Test by file:
$ python test_file.py -s silva -f example.txt
or:
$ python test_file.py -s ceci -f example.txt
- Marquiafavel, V.; Bokan, A. and Zavaglia, C. (2014). "PETRUS: A rule-based grapheme-to-phone converter for Brazilian Portuguese". In: J. Baptista et al. (Eds.): PROPOR 2014, LNAI 8776, Springer, Heidelberg (2014).
- Cristófaro-Silva, T. (2000). "Fonética e fonologia dos português: roteiro de estudos e guia de exercícios". 3a ed., São Paulo: Contexto.
- Cagliari, L. (2009). "Elementos de fonética do português brasileiro". São Paulo: Paulistana.
- Silva, D. (2011). "Algoritmos de processamento da linguagem e síntese de voz com emoções aplicados a um conversor texto-fala baseado em HMM". Tese de Doutorado. Programa de Pós-Graduação em Engenharia Elétrica, COPPE, Universidade Federal do Rio de Janeiro, RJ, 2011