Advances in multimedia device processing capabilities and the Internet have contributed to the proliferation of collaborative applications such as distance learning (DL) webcasting. The challenge that remains to be addressed in existing DL systems is providing access to multimedia content to any user regardless of device capabilities, IP network heterogeneity, and personal preferences such as modality. A novel metadata-driven adaptation architecture is presented to overcome these challenges by performing (1) application layer transcoding that adapts the presentation format of DL content to match device media decoding capabilities and user desired modality, (2) bitstream transcoding to adapt multimedia to match device processing capabilities and encoding bit-rate supported by the network. Results demonstrate how the proposed system delivers personalized DL content with negligible transcoding overhead while meeting end-user environmental restrictions.