This is an implementation of the bidirectional LSTM network for speech recognition as defined by Graves & Schmidhuber (2005). It's defined in the Flux machine learning library for the Julia programming language.
Graves, A., & Schmidhuber, J. (2005). Framewise phoneme classification with bidirectional LSTM and other neural network architectures. Neural Networks, 18(5-6), 602-610.