The goal of combining the robustness of neural networks and the expressivity\nof symbolic methods has rekindled the interest in neuro-symbolic AI. Recent\nadvancements in neuro-symbolic AI often consider specifically-tailored\narchitectures consisting of disjoint neural and symbolic components, and thus\ndo not exhibit desired gains that can be achieved by integrating them into a\nunifying framework. We introduce SLASH -- a novel deep probabilistic\nprogramming language (DPPL). At its core, SLASH consists of\nNeural-Probabilistic Predicates (NPPs) and logical programs which are united\nvia answer set programming. The probability estimates resulting from NPPs act\nas the binding element between the logical program and raw input data, thereby\nallowing SLASH to answer task-dependent logical queries. This allows SLASH to\nelegantly integrate the symbolic and neural components in a unified framework.\nWe evaluate SLASH on the benchmark data of MNIST addition as well as novel\ntasks for DPPLs such as missing data prediction and set prediction with\nstate-of-the-art performance, thereby showing the effectiveness and generality\nof our method.\n