An encoding from base64 to consonant/vowel syllables
Base64 is an encoding that is capable of converting binary data to printable characters. It uses a set of 64 characters: lower case letters, upper case letters, digits, plus (+), slash (/) and equality (=) as a 65th for a possible padding at the end.
However, Base64 can be difficult to read, since one might confuse 0 and O, or I and l and 1.
Sometimes, you have to type or speak out Base64 encoded strings, for example when entering a one time password sent to you via mail.
What if c0JpKL1n would be replaced by fiso-neja-nino-suho? Wouldn't that be much easier to read and to type and to memorize (if it was a password)? It would.
The alphabet consist of 5 vowels and 21 consonants. If we omit consonants that sound similar and keep 13 of them, we have 13 * 5 possible syllables consisting of a consonant and a vowel (hence cv-encoding). We can now replace each character of Base64 with a syllable. To increase readability, we add a dash or a space after every two syllables.
Which consonants do we pick? We want to avoid homophones, that is syllables that sound the same but are written differently (e.g. ji and gi). We want to avoid sets of consonants that can be pronounced differently (e.g. in German, the v can be pronounced as both f and w, we should only use one consonant per set).
The problem is that these phonemes differ between languages. The best I came up with - having had English and German in mind - is bdfghklmnprst.