@User binary raw files are extremey simple to process, i guess also from python. basically there is a ASCII header section showing the number of variables (nvars), the list of signals names saved and the number of timesteps (npoints) . The ascii header is followed by a binary blob that is an array of (nvars * npoints) doubles. in xschem this simple C program reads a raw file and creates a vcd file. The function that reads the binary blob (after getting npoints and nvars from he ascii header) is read_binary_block(), it is just 10 lines of code. Of course if raw file is really big and the whole dataset does not fit into RAM you will probably need to filter data in the program somehow.