then it makes sense to store, along with the data, two codecs that will make a array of bitmaps and an uncompressed raw audio file. (project gutenberg, for example, always has a plain text version whenever there's a more elaborate format.)

That's with historical proprietary formats.