This paper describes how digital talking books (DTBs) with embedded functionality for learners can be generated from content structured according to the OU OpenLearn schema. It includes examples showing how a software transformation developed from open source components can be used to remix OpenLearn content, and discusses issues concerning the generation of synthesised speech for educational purposes. Factors which may affect the quality of a learner's experience with open educational audio resources are identified, and in conclusion plans for testing the effect of these factors are outlined. (Contains 20 footnotes, 1 table and 6 figures.)