This presentation will introduce libsigmf, a FOSS library providing C++ functionality and utilities for generating, using, and translating digital recordings of signals using the Signal Metadata Format (SigMF), itself a FOSS effort that was first publicly announced at FOSDEM'17. This talk will provide an introduction to SigMF and its design philosophy, sharing its strengths, current state, and intended usage patterns. It will then present details on libsigmf, which can be integrated and used in other projects, like GNU Radio, or directly via a C++ API. Finally, we will discuss the use of SigMF in machine learning, an application for which it is particularly well suited. Additionally, the 'Breakthrough Listen' project at the UC Berkeley SETI Research Center will share their use of libsigmf and the availability of SigMF datasets from 'Breakthrough Listen' radio telescope facilities.