Convert UTF-16 to UTF-8

Two consecutive integers (0-255) form one UTF-16 code unit. The byte order setting controls which byte is high. In Big-endian: 0 72 = (0<<8)|72 = U+0048 = “H”. In Little-endian: 72 0 = (0<<8)|72 = U+0048 = “H”. A common confusion: many sources write 72 0 101 0 as “byte pairs for ‘He'” – that’s actually little-endian. With Big-endian selected, you need 0 72 0 101.

High surrogates (U+D800-U+DBFF) must be followed by low surrogates (U+DC00-U+DFFF). The decoder combines them via 0x10000 + ((high - 0xD800) << 10) + (low - 0xDC00). Unlike many converters, this tool raises an explicit error on lone surrogates rather than silently emitting U+FFFD – so silently broken inputs are visible, not papered over.

Byte Order Mark – the codepoint U+FEFF. As UTF-16, it serializes as FE FF (BE) or FF FE (LE), letting readers detect the byte order. The decoder auto-strips a leading FEFF/FFFE if present (stats will say “BOM stripped”). On reverse, toggle the BOM checkbox to prepend it.

For hex, it IS byte-swapping per 16-bit unit: BE 0048 → LE 4800. For decimal byte-pairs, the swap happens at the pair level (order of the two bytes flips). For binary, the high and low 8-bit halves swap. The decoder applies the right transform per format automatically.

Non-BMP characters (emoji, rare scripts) use surrogate pairs – TWO 16-bit code units in UTF-16. Example: 🌍 (U+1F30D) is D83C DF0D. The grid shows these as “pair (2)” so you can tell where surrogate-pair-driven non-BMP characters are.

UTF-8 uses 1 byte for ASCII (U+0000-U+007F), 2 for Latin/Cyrillic/Greek/Hebrew/Arabic (U+0080-U+07FF), 3 for most CJK and the rest of BMP (U+0800-U+FFFF), 4 for non-BMP (U+10000-U+10FFFF). The stats line shows the precise count.

The browser’s TextDecoder('utf-16be') works for raw byte buffers but silently substitutes U+FFFD for invalid surrogates. We do it manually so we can report which exact code unit position caused the failure – useful when debugging real-world UTF-16 streams.

No. Parsing, decoding, and UTF-8 encoding happen entirely in your browser. No network requests.

200,000 characters. The lower cap keeps the UI responsive.

UTF-8 dominates web, files, APIs, and modern protocols. UTF-16 is the internal string representation in JavaScript engines, Windows APIs (UCS-2 originally), and Java char. If you’re storing or transmitting text, almost always UTF-8. If you’re poking at JS string code units (.charCodeAt) or Windows wide-char APIs, you’re touching UTF-16.