UTF-16 to Base64: Encode UTF-16LE or UTF-16BE Text

How to encode text as UTF-16 Base64

1. Type or paste your text into Input · text. Charset is preset to UTF-16LE.
2. Copy the Base64 from Output · Base64. The Input figure shows the UTF-16 byte count, twice the length of plain English text.
3. For a big-endian consumer, switch Charset to UTF-16BE.
4. Swap decodes the result again for a quick check; Base64 to UTF-16 is the dedicated decoder.

UTF-16LE vs UTF-16BE

UTF-16 stores each character as one or two 16-bit code units. Little-endian writes the low byte first, big-endian the high byte first. Windows and .NET use little-endian; Java’s UTF-16 encoder and many network formats use big-endian.

Characters outside the Basic Multilingual Plane, including every emoji, take two code units called a surrogate pair. 👋 (U+1F44B) becomes D83D DC4B, which is the 4 bytes in the last row.

Since every ASCII letter is paired with a zero byte, UTF-16 Base64 of English text is dotted with A characters. An unknown string with that pattern is worth decoding as UTF-16.

Where UTF-16 Base64 is used

• PowerShell -EncodedCommand. powershell -EncodedCommand RwBlAHQALQBEAGEAdABlAA== runs Get-Date. A frequent reason such commands fail is a script encoded as UTF-8. The PowerShell EncodedCommand tool adds the command prefix, warns about the length limit of cmd.exe and also decodes commands found in logs.
• .NET code. Encoding.Unicode in C# is UTF-16LE, so Convert.ToBase64String(Encoding.Unicode.GetBytes(s)) produces the same output as this page.
• Windows data. Strings from the registry, from NVARCHAR columns cast to binary in SQL Server and from many Windows APIs are UTF-16LE.

Byte order marks

This tool writes no byte order mark, which is what PowerShell and most APIs want. Some libraries add one automatically:

If your Base64 must match one of those, encode with an explicit byte order in code, as below.

UTF-16 to Base64 in code

import base64

base64.b64encode("Get-Date".encode("utf-16-le")).decode()   # 'RwBlAHQALQBEAGEAdABlAA=='
base64.b64encode("Hi".encode("utf-16-be")).decode()         # 'AEgAaQ=='

// Node.js has a little-endian codec only
Buffer.from("Get-Date", "utf16le").toString("base64");   // "RwBlAHQALQBEAGEAdABlAA=="

[Convert]::ToBase64String([Text.Encoding]::Unicode.GetBytes('Get-Date'))

Use utf-16-le rather than utf-16 in Python unless you want the BOM. More examples: Python, Node.js.

FAQ