About

Base32 encoding represents binary data using a 32-character alphabet, producing output that is case-insensitive and safe for case-folding filesystems, DNS labels, and human verbal communication. RFC 4648 defines the canonical alphabet (A - Z, 2 - 7) with = padding. Each input byte (8 bits) maps to 85 output characters, yielding a 160% size expansion. Incorrect variant selection or stray characters cause silent data corruption. This tool validates the input alphabet, strips whitespace, and decodes the bitstream faithfully.

Four variants are supported. Crockford excludes I, L, O, U to avoid human misreading. z-base-32 reorders characters to place the most visually distinct glyphs at the highest-frequency bit patterns. Base32Hex preserves sort order by using 0 - 9 then A - V. Mixing alphabets between variants produces garbage output without any error signal from a naive decoder. This converter detects alphabet mismatches and warns before producing output.

Formulas

Base32 encoding maps every 5 bits of input to one character from a 32-symbol alphabet. Decoding reverses this: each character yields a 5-bit value, the bits are concatenated, and the resulting bitstream is split into 8-bit bytes.

n_chars = n_bytes × 85

The total output length is rounded up to the nearest multiple of 8 with = padding (in RFC 4648 and Base32Hex):

n_padded = 8 × ceil(n_chars8)

Decoding algorithm per character c_i:

bits = bits + pad(toBinary(lookup(c_i)), 5)

Then extract bytes:

byte_j = bits[j × 8 .. (j + 1) × 8]

Where n_bytes is the number of input bytes, n_chars is the number of Base32 characters (excluding padding), lookup maps a character to its 0 - 31 integer value using the selected variant's alphabet, and pad left-pads the binary string to 5 digits.

Reference Data

Variant	Alphabet	Padding	Case Sensitive	Standard	Use Case
RFC 4648 (Standard)	A - Z, 2 - 7	=	No	RFC 4648 §6	SASL, XMPP, general encoding
Base32Hex	0 - 9, A - V	=	No	RFC 4648 §7	Sort-preserving encoding, NSEC3
Crockford	0 - 9, A - HJ - KM - NP - TV - W	None	No	Crockford spec	Human-readable IDs, check symbols
z-base-32	ybndrfg8ejkmcpqxot1uwisza345h769	None	Yes (lower)	z-base-32 spec	Mnet-URI, human-oriented hashes
Encoding Size Comparison (per input length)
Input Bytes	Base32 Chars	Padding Chars	Total Output	Bit Waste	Expansion Ratio
1	2	6	8	2 bits	800%
2	4	4	8	4 bits	400%
3	5	3	8	1 bit	267%
4	7	1	8	3 bits	200%
5	8	0	8	0 bits	160%
Character Encoding Reference
Value 0	RFC: A		Hex: 0	Crockford: 0	z-base: y
Value 9	RFC: J		Hex: 9	Crockford: 9	z-base: e
Value 15	RFC: P		Hex: F	Crockford: F	z-base: x
Value 25	RFC: Z		Hex: P	Crockford: P	z-base: 3
Value 31	RFC: 7		Hex: V	Crockford: W	z-base: 9

Frequently Asked Questions

You get corrupted output without any error. For example, the Crockford alphabet maps character I to 1 and O to 0, while RFC 4648 treats I as value 8. A single character mismatch shifts 5 bits, cascading errors through every subsequent byte. Always verify the source system's variant before decoding.

Base32 uses 5 bits per character versus Base64's 6 bits. For n input bytes, Base32 produces ceil(n × 8 ÷ 5) characters (160% expansion) while Base64 produces ceil(n × 8 ÷ 6) characters (133% expansion). The tradeoff is case insensitivity and better human readability.

Padding ensures output length is always a multiple of 8. Since 5 input bytes map perfectly to 8 Base32 characters, any input not divisible by 5 requires padding. Specifically: 1 byte → 6 pads, 2 bytes → 4 pads, 3 bytes → 3 pads, 4 bytes → 1 pad. Crockford and z-base-32 omit padding entirely.

Yes. Base32 encodes arbitrary binary data. When decoding such data back, the result is raw bytes. If those bytes are not valid UTF-8, this converter will show replacement characters (U+FFFD) or display the hex dump instead. For binary payloads, use the hex output mode or download the result as a raw file.

The converter processes inputs up to 5 MB of encoded text. Beyond this, browser memory constraints make in-page processing unreliable. For a 5 MB Base32 string, the decoded output is approximately 3.125 MB (ratio of 5 ÷ 8). Processing time is under 200 ms for typical inputs.

Crockford's specification defines an optional check symbol appended as the last character. The check value is the remainder of dividing the numeric value by 37, mapped to symbols 0 - 9, A - Z, *, ~, $, =, U. This converter strips check symbols during decode. Enable the check-symbol option to validate before stripping.