In 2004, the Berlin Database of Emotional Speech (Berlin EmoDB) has made publicly available, and since then, it has been utilized in numerous studies on emotional speech, with over 3,000 citations. We now extend this database with original material that was previously absent from the distributed versions within the research community. This includes ambiguous samples that were not agreed upon by at least 80 % of the raters, the addition of glottograms for all samples, and a new rater label indicating the perceived naturalness of the samples. The paper provides detailed descriptions and reports on preliminary studies conducted using this extended data. For instance, incorporating glottogram information has been shown to improve the average recall of an SVM classifier by 8.1 % UAR.