Convert UTF-8 to UTF-16

Both encode the same Unicode codepoints, just differently. UTF-8 uses 1-4 bytes per codepoint with variable-width sequences. UTF-16 uses 2 or 4 bytes (one or two 16-bit code units, surrogate pair for non-BMP). They roundtrip losslessly through the codepoint layer.

Non-BMP characters (U+10000+) don’t fit in one 16-bit code unit, so UTF-16 splits them into a high surrogate (U+D800-U+DBFF) + low surrogate (U+DC00-U+DFFF). The encoding: high = 0xD800 + ((cp – 0x10000) >> 10); low = 0xDC00 + ((cp – 0x10000) & 0x3FF). The grid labels these as “pair (2)” so you see them.

Every high surrogate must be immediately followed by a low surrogate; every low surrogate must follow a high surrogate. Violations throw explicit errors with position – unlike some converters that silently substitute U+FFFD, this tool surfaces broken streams so you can fix them.

U+FEFF prepended. Serializes as FE FF in BE or FF FE in LE – a reader can detect byte order from these. Distinct from the UTF-32 BOM (4 bytes 00 00 FE FF). The decoder auto-strips on detect.

Historical: when ECMAScript 1 was specified (1997), Unicode fit in 16 bits (BMP only). JS adopted “UCS-2” code units; later when Unicode expanded past U+FFFF, surrogate pairs were retrofitted into UTF-16 to maintain backwards compatibility. str.length returns code units, NOT codepoints – so "🌍".length is 2.

For ASCII: UTF-16 is 2× UTF-8 (every ASCII char becomes 2 bytes vs 1). For mixed Latin: roughly even. For CJK: UTF-16 is ~33% smaller (2 bytes per char vs 3 in UTF-8). For emoji: same (4 bytes per char in both, just different splits).

No – that would be UCS-2, which is a strict subset of UTF-16 that rejects non-BMP. Modern UTF-16 always supports surrogate pairs. If your downstream consumer is strict UCS-2, codepoints beyond U+FFFF will fail there, not in this tool.

No. The conversion runs entirely in your browser – nothing is sent to a server, logged, or stored, and the tool keeps working offline once the page has loaded.

200,000 characters per run, roughly a short novel chapter. The cap keeps the browser responsive; for bigger jobs, split the text and run it in parts.

Inverse direction. That tool parses UTF-16 code units and produces text + UTF-8 bytes. This tool starts from UTF-8 text and produces UTF-16 code units. Both validate surrogates strictly.