MP3 to AU Converter

How to Convert MP3 to AU Online

1. Upload Your MP3 Files: Drag and drop one or more .mp3 files into the dropzone, or click "Add Files" to browse. Batch conversion is supported — every file uses the same settings.
2. Pick a Quality Preset: Default is Very High (Recommended). Choose Highest for archival output, Medium or Low to shrink size, or Lowest for telephony-grade speech where bandwidth is more important than fidelity.
3. Set Audio Channel and Sample Rate (Optional): Audio Channel offers Original, Mono, or Stereo — pick Mono for IVR / phone systems. Audio Sample Rate exposes 8000 Hz, 12000 Hz, 16000 Hz, 24000 Hz, 44100 Hz, and 48000 Hz; 8000 Hz mono matches the classic Sun /dev/audio and telephony pipeline, while 44100 Hz preserves the MP3's source rate.
4. Trim and Convert (Optional): Use the Trim controls to set a start time and duration in seconds or HH:MM:SS.sss if you only need a clip. Click Convert and download the .au files. Processing runs in your browser session — no sign-up, no watermark, files stay private.

Why Convert MP3 to AU?

AU (also written .au or .snd) is an audio container introduced by Sun Microsystems and adopted as the native sound format on NeXT systems. Each file starts with a 24-byte big-endian header — magic number 0x2e736e64 (".snd" in ASCII), data offset, data size, encoding ID, sample rate, and channel count — followed by raw audio samples. That trivial structure is why AU survived long after Sun did: anything that can read a 6-word header can decode it.

You'd convert MP3 → AU when a downstream system requires the AU container, not because AU sounds better (MP3 is lossy, so re-encoding to PCM AU gives you a much larger file with no recovered detail).

• Java Sound API playback — javax.sound.sampled natively recognises AU alongside WAV and AIFF. AU is the simplest format to ship inside a JAR for Clip or AudioInputStream playback in desktop Swing/JavaFX apps without pulling in a third-party decoder.
• Telephony, IVR, and PBX prompts — Asterisk, FreeSWITCH, and many legacy IVR stacks accept AU at 8000 Hz mono with μ-law encoding (the North-American/Japanese G.711 variant) or A-law (the European G.711 variant). That's the same encoding ITU-T G.711 standardised in 1972 for digital phone trunks.
• Unix /dev/audio and SPARCstation heritage — On classic Solaris and SunOS, piping a μ-law AU file straight to /dev/audio plays it. Scientific computing labs running legacy Sun pipelines still depend on this.
• Embedded and DSP toolchains — MATLAB's audioread, SciPy's scipy.io ecosystem, and many C reference players read AU with no library dependency, which suits firmware test fixtures and signal-processing course material.
• Archival of speech-only content — For voicemail-style or call-centre recordings where 8 kHz μ-law is "good enough", AU is a smaller, simpler alternative to WAV with the same lossless characteristics at that bit depth.

MP3 vs AU — Format Comparison

AU Encoding Quick Guide

The AU spec defines 28 encoding IDs. These are the ones you actually hit when converting from MP3:

Quality presets in xconvert's encoder map onto these output codecs and bitrates; the higher the preset, the higher the bit depth and sample rate retained.

For the reverse direction, see AU to MP3. For lossless masters that browsers actually play, MP3 to FLAC or MP3 to WAV are usually a better fit. Need to shrink an oversized MP3 first? Try Compress MP3 before converting.

Frequently Asked Questions

Will converting MP3 to AU improve audio quality?

No. MP3 is a lossy format — the perceptual encoder permanently discarded data when the original was encoded. Converting to AU PCM produces a larger file that contains exactly the same audible information as the MP3, just stored uncompressed. Use AU when a system requires the format, not as a quality upgrade.

Why is the AU file roughly 10× larger than my MP3?

AU normally stores linear PCM (uncompressed). A 1-minute stereo MP3 at 128 kbps is about 1 MB; the same audio as 16-bit PCM AU at 44.1 kHz stereo is roughly 10 MB (44,100 samples/sec × 2 bytes × 2 channels × 60 seconds ≈ 10.1 MB). That ratio is mathematical, not a tool quirk.

Should I pick μ-law or A-law for telephony?

μ-law is the G.711 variant used on North-American and Japanese phone networks; A-law is the G.711 variant used across Europe and the rest of the world. Both compress 14-bit dynamic range into 8 bits using a logarithmic curve. Match the platform you're targeting — Asterisk, FreeSWITCH, and most US-built IVRs default to μ-law, so set Audio Channel = Mono and Audio Sample Rate = 8000 Hz, then pick the μ-law output for that pipeline.

Is AU the same as .snd?

Functionally yes — the magic number at the start of every AU file is the four ASCII bytes .snd (hex 0x2e736e64). NeXT systems used the .snd extension; Sun and most modern tooling use .au. The header layout, byte order, and encoding IDs are identical, so renaming .au to .snd produces a valid file that any AU-aware reader will accept.

Why does my AU file fail to play in a web browser?

No major browser ships native AU support in the HTML5 <audio> element — Chrome, Firefox, Safari, and Edge will not decode it. AU's home is server-side audio, Java desktop apps, and Unix scripts. If you need browser playback, convert MP3 to a web-friendly format like MP3 to WAV (universally supported) or keep the MP3 itself.

Will the AU output preserve my MP3's bitrate or sample rate?

By default the converter resamples to the Quality Preset's target. To keep the source intact, leave Audio Sample Rate on Original and Audio Channel on Original — the encoder will keep the MP3's decoded sample rate (typically 44100 Hz) and channel count. The AU output is uncompressed PCM, so its byte rate is fixed by sample rate × bit depth × channels, regardless of the MP3's input bitrate.

Can Java's javax.sound.sampled play the file directly?

Yes for the common encodings. AudioSystem.getAudioInputStream(file) reads AU containing PCM, μ-law, or A-law without an SPI plug-in — those AudioFormat.Encoding values are built in. Some sound cards refuse to play 8-bit 8 kHz μ-law directly; if you hit that, call AudioSystem.getAudioInputStream(targetFormat, sourceStream) to up-convert to 16-bit PCM at runtime.

Why is AU a 24-byte header — is there a longer variant?

The minimum header is six 32-bit big-endian words: magic, data offset, data size, encoding, sample rate, channels — 24 bytes total. The data-offset field can point further into the file to leave room for an optional NUL-terminated annotation block (textual metadata). Most encoders write 28 bytes (24 + 4 of annotation padding), but readers should always trust the data-offset value rather than assuming a fixed 24.

Can I batch convert and trim in the same job?

Yes. Drop multiple MP3 files in at once — every file gets the Quality Preset, Audio Channel, Sample Rate, and Trim window you set. If you only need to trim without changing format, use the dedicated Audio Trimmer; for any-to-AU conversion from a non-MP3 source, use the general Audio to AU tool.