4 ms·

This dataset has been a huge boon for my English acoustic models that recognize many accents at once.