Towards a systematic and quantitative analysis of vocal tract data

Articulatory data can nowadays be obtained using a wide range of techniques, such as real-time magnetic resonance (RT-MRI), enabling acquisitions of large amounts of data. A major challenge arises: analysing these new large data sets to extract meaningful information regarding speech production in an expedite and replicable way. Traditional approaches such as superimposing vocal tract profiles and qualitatively characterizing relevant properties and differences, although providing valuable information, are rather inefficient and subjective. Therefore, analysis must evolve towards a more automated, quantitative approach. To tackle this issue we propose the use of objective measures to compare the configurations assumed by the vocal tract during the production of different sounds. The proposed framework provides quantitative data regarding differences pertaining meaningful regions under the influence of various articulators. Visual representation of such data is a key part of the proposal and some concrete forms of visualization are proposed to depict the differences found and corresponding direction of change. Application examples concerning the articulatory characterization of EP vowels are presented with promising results, paving the way towards automated and objective analyses of articulatory data.