However, different speech-to-text programs have different levels of ability and complexity, with some using advanced machine learning to constantly correct errors flagged up by users so that they are not repeated. Others are downloadable software which is only as good as its latest update.