This dataset contains direct assessment data collected and curated during machine translation dataset creation and model evaluation for Mozambican languages. The data is intended for evaluating the quality and accuracy of translation and speech models.