Binary text starts with an encoding choice
Binary describes a number system, but a message needs another rule: a character encoding. The letter A becomes the byte 65 in both ASCII and UTF-8, written as 01000001 in binary. That simple example can hide an important detail. Many characters need more than one UTF-8 byte, so counting the groups is not the same as counting the letters on screen.
Choose strict ASCII when checking a system that accepts only values from zero to 127. The tool rejects a character outside that range. UTF-8 is the useful choice for ordinary multilingual documents. A Hindi word, an accented surname or an emoji may occupy several bytes while remaining a short piece of readable text. Encoding these characters does not translate their language.
When decoding, this translator accepts compact binary runs and complete byte groups separated by whitespace, commas or colons. Each run must contain a multiple of eight bits. It will not join two broken groups and pretend they were a valid byte. A syntactically correct byte sequence can still be invalid UTF-8; that produces a separate error with the original input left available.
For a batch, every file uses the selected settings. A failed file does not remove successful outputs, and the ZIP report identifies failures by selection order. Read the decoded text before passing it into a document workflow. Binary encoding is a representation, not encryption: anyone with the same encoding rule can reverse it.