Proceedings of the Xth Conference, Helsinki 2008
Edited By Martti Vainio, Reijo Aulanko and Olli Aaltonen
16. CriaExperimento: obtaining data about speech perception from internet 159
CRIAEXPERIMENTO: OBTAINING DATA ABOUT SPEECH PERCEPTION FROM INTERNET Ana Cristina Fricke Matte Cecilio Cosac Fráguas Rubens Takiguti Ribeiro 1 Introduction Some of the resolutions to phonetic questions are so exact and, at the same time, so big that they cannot be reached without a very large corpus. One such question was the main reason that made us design the CriaExperimento (http://www. sourceforge.net/projects/criaexperimento): we were (and we still are) concerned about the relationship between the automatically detected end of the stress group and the perception of speech emphasis. This problem is important in order to check the possibility of using this stress group as a prosodic sentence in automatic speech analysis. However, we knew that it was not possible to say anything about that with only a small corpus of speech perception. It was necessary, in order to check our hypothesis, to get data where, at least the following characteristics of the informants were controlled: – gender – scholar grade – age – geographic region (in Brazil, because of its wideness, this is extremely relevant) Beyond that, the acoustic parameters involved also enlarge the data needed, because of their number: pitch, intensity and duration, at least. CriaExperimento is not a program for obtaining speech data, but for obtaining data about the perception of speech. The acoustic parameters are obtained from the previously registered sounds, the perception of which is tested by Cria- Experimento. The web is a great way for obtaining a large corpus: the experiment can be done anytime, by anyone, anywhere....
You are not authenticated to view the full text of this chapter or article.
This site requires a subscription or purchase to access the full text of books or journals.
Do you have any questions? Contact us.Or login to access all content.