Base91 Encoding

Encryption uses a method similar to base64 but with an expanded alphabet to 91 characters: the 94 printable ASCII characters (from 0x21 to 0x7E) omitting -: dash (0x2D), \: backslash (0x5C) and ': apostrophe (0x27).

To simplify, the data is divided into 13-bit binary packets (ie 2 ^ 13 = 8192 values) which are then encoded in 2 letters of the alphabet (which contains 91 characters and 91 ^ 2 = 8281).

Example: DCODE is coded xJ,Jc,A in BasE91

The decryption performs the inverse conversion to base 91.

To simplify, the message is split into groups of 2 letters which are converted into 13-bit data. The message is then read with the original encoding (ASCII or Unicode).

Example: The message encoded in Base 91 xJ,Jc,A is decoded in DCODE.

The message is composed only of printable ASCII characters (except dash, backslash and apostrophe) that is 91 characters maximum.

Base91 encoded messages are on average 23% longer than the original message.

The author writes his code basE91 (with a capital E)

Joachim Henke claim it was invented in 2005, see here (link)