Teflon Pronunciation Assessment Challenge

<h1>Speech data for the Teflon Pronunciation Assessment Challenge.</h1> <p>This dataset contains recordings of isolated words in Norwegian spoken by children in the age 4-12. Some of the children are motherthongue Norwegian, others are immigrants with different linguisti backgrounds. The fine names were randomized to hide the identity of the speakers. The pronunciation of each utterance has been assessed by at least one assessor on a scale 1-5 with the following meaning:</p> <ol> <li>Not at all identifiable as the target word</li> <li>Difficult to identify as the target word</li> <li>Slight phonemic error(s)</li> <li>Subphonemic error(s) or "unexpected variants"</li> <li>Prototypical, adult-like</li> </ol> <p>The data is divided into train and test. The corresponding CSV files contain, for each utterance, information about the spoken word and the score. Scores are only given for the training data.</p>