Batch Soft-DTW(Dynamic Time Warping) in TensorFlow2 including forward and backward computation

Custom TensorFlow2 implementations of forward and backward computation of soft-DTW(Dynamic Time Warping) algorithm in batch mode, which is proposed in paper 《Soft-DTW: a Differentiable Loss Function for Time-Series》.

I have implemented two versions of soft-DTW, one is the original paper, the other is Parallel Tacotron2's paper(with warp penalty). For latter version, I solved the equations of backward computation myself.

If you have questions or improvements about the code, welcome to submit issues ASAP!