Base64 to Audio File
Paste a Base64 string or a data:audio/…;base64,… URI and get the audio file back. The container is detected from the decoded bytes, so the result carries the right MIME type and extension — WAV, MP3, Ogg, FLAC, M4A, AAC, WebM, AIFF, AMR, WMA or MIDI. Nothing is re-encoded and nothing is uploaded: the decode runs in your browser via WebAssembly.
About this tool
Audio travels through APIs, JSON payloads, databases and HTML attributes as Base64 text.
This tool turns that text back into the file it came from: paste the Base64 (or the whole
data:audio/…;base64,… URI), and the decoded bytes come back as a data: URL you can play,
save, or paste into the address bar.
Nothing is re-encoded. The bytes that come out are byte-for-byte the bytes that were encoded — the work is the tolerant cleanup of the Base64 text plus reading the decoded file's magic header so the result gets the correct MIME type and extension.
What gets detected
With Format left on auto, the first bytes of the decoded payload decide the container:
| Container | Header it is recognized by | Result |
|---|---|---|
| WAV | RIFF … WAVE | audio/wav · .wav |
| MP3 | an ID3 tag, or a bare MPEG frame sync | audio/mpeg · .mp3 |
| Ogg (Vorbis/Opus/FLAC) | OggS | audio/ogg · .ogg |
| FLAC (native) | fLaC | audio/flac · .flac |
| MP4 / M4A | ftyp at offset 4 | audio/mp4 · .m4a |
| AAC (ADTS) | a 0xFFF… ADTS sync word | audio/aac · .aac |
| WebM | the 1A 45 DF A3 EBML marker | audio/webm · .webm |
| AIFF / AIFC | FORM … AIFF | audio/aiff · .aiff |
| AMR | #!AMR | audio/amr · .amr |
| WMA / ASF | the ASF header GUID | audio/x-ms-wma · .wma |
| MIDI | MThd (or a RIFF … RMID) | audio/midi · .mid |
Input the decoder forgives
- Line breaks, spaces and tabs anywhere in the payload.
- A single layer of wrapping
"or'quotes, as pasted out of JSON or a shell command. - A
data:URI prefix — its declared MIME type is reported when it disagrees with the bytes, but the sniffed bytes win. - The URL-safe alphabet (
-and_instead of+and/). - Missing
=padding.
Worked example
Input (truncated):
UklGRuwAAABXQVZFZm10IBAAAAABAAEAQB8AAEAfAAABAAgAZGF0YcgAAACAob/W…
With File name beep, Format auto and strict checking on, the result is a
data:audio/wav;base64,… URL. The chat and CLI surfaces return the same bytes as a real
download named beep.wav; gizza tool base64-to-audio-file data=… filename=beep --out beep.wav
writes the file directly.
Limits & edge cases
- Payloads that decode to more than 32 MiB are rejected rather than truncated.
- Reject bytes that aren't audio (strict, on by default) applies only when Format is
auto. A payload that decodes cleanly but carries, say, a PNG header is refused with the type it actually looks like. Turn it off to save the bytes asapplication/octet-streaminstead. - Choosing a Format explicitly always wins — use it for headerless payloads such as raw AAC or a bare PCM dump, where there is no header to sniff. The summary notes when the forced format disagrees with what the bytes look like.
- A file name is reduced to a bare stem: directories are stripped, unusual characters become
-, and any extension you type is replaced by the one the resolved format implies. - Only
data:URIs with;base64are accepted; percent-encodeddata:URIs are not audio in practice and are rejected with a clear message.
FAQ
Does this convert between audio formats, for example Base64 MP3 to WAV?
No. This tool only reverses the Base64 encoding — the decoded bytes are exactly what was encoded, so a Base64 MP3 comes back as an MP3. Setting Format changes the MIME type and file extension that get attached, not the audio data itself. To actually re-encode between codecs, use an audio conversion tool that runs a real encoder.
My Base64 decodes but the tool says it isn't audio. What now?
The bytes carry no header this tool recognizes. The error names what they look like instead (a PNG, a PDF, plain text, and so on), which usually means the wrong field was copied. If you know the payload really is audio but has no container header — raw AAC frames or a headerless PCM dump — set Format to that container explicitly, or turn off Reject bytes that aren't audio to save the bytes as-is.
Can I paste a whole data: URI, or must I strip the prefix first?
Paste the whole thing. A leading data:audio/mpeg;base64, (or any other media type) is
stripped for you. Its declared type is only a hint: if it disagrees with the decoded bytes,
the summary says so and the bytes decide the extension — that mismatch is exactly why a file
saved from a data: URI sometimes ends up with the wrong suffix.
Why does my Base64 fail with "invalid Base64" when it looks fine?
Something in the payload is not a Base64 character. The error names the offending character
and its position, which is usually a stray <, a %, a \n escape that stayed literal, or
HTML entities from a copy out of a rendered page. Whitespace, quotes, the URL-safe -_
alphabet and missing = padding are all handled already, so anything else that trips the
decoder is genuinely not part of the data.
Is my audio uploaded anywhere?
No. The page runs the decoder as WebAssembly inside your browser, and the command-line tool runs it locally. The Base64 you paste never leaves the device.
How large a payload can I decode?
Up to 32 MiB of decoded audio, which is roughly 43 MiB of Base64 text. Larger payloads are rejected with a clear message rather than silently truncated — a half-decoded audio file is worse than none.
Developer & Automation Access
Run it from the terminal
Same engine as this page, headless — via the gizza CLI:
gizza tool base64-to-audio-file "UklGRuwAAABXQVZFZm10IBAAAAABAAEAQB8AAEAfAAABAAgAZGF0YcgAAACAob/W…"New to the CLI? Get gizza →
Open it by URL
Pre-fill and auto-run this tool with query parameters — the names match the API/CLI:
https://gizza.ai/tools/base64-to-audio-file/?data=UklGRuwAAABXQVZFZm10IBAAAAABAAEAQB8AAEAfAAABAAgAZGF0YcgAAACAob%2FW%E2%80%A6&filename=audio&format=auto&strict=trueMachine-readable descriptor: tool.json — title + parameters JSON Schema for agents.