Trim AU

How to Trim AU Audio Online

1. Upload Your AU File: Drag and drop, or click "+ Add Files" to select your AU (Sun/NeXT) audio file. Files stay in your browser session — no account, no watermark. Batch trimming is supported, so you can queue multiple AU clips.
2. Set Start Time and Duration: Under "Trim", enter the start position (e.g. 00:00:05) and the length you want to keep (e.g. 00:00:30). Times are in hh:mm:ss format. Anything outside the window is discarded.
3. Adjust File Compression, Sample Rate, and Channels (Optional): Under "File Compression", pick a Quality Preset (Highest to Lowest) or switch to Constant Bitrate for a fixed kbps target. Change the Audio Sample Rate (8000 Hz for telephony, 22050/44100/48000 Hz for music) and the Audio Channel (mono or stereo) if you need to downmix or upmix.
4. Trim and Download: Click "Trim". The trimmed segment is re-encoded to AU and saved to your downloads — same .au extension, same .snd magic header.

Why Trim AU Files?

The AU format was introduced by Sun Microsystems and became the de-facto sound format on early SunOS/Solaris workstations, NeXT computers, and the first wave of Java applets. Because Java 1.0/1.1's sun.audio.AudioPlayer and the early AudioClip API only accepted 8-bit μ-law AU at 8 kHz, a generation of Unix and Web audio assets was archived in this exact shape. Trimming AU directly — instead of round-tripping through MP3 or WAV — preserves the original encoding, sample rate, and the minimal 24-byte .snd header that legacy tooling expects.

• Edit telephony recordings without quality loss — Voicemail boxes, IVR prompts, and call-recording systems on Solaris and AIX still emit 8 kHz μ-law AU. Trimming as AU keeps the μ-law companding intact instead of decoding to PCM and re-encoding (which can introduce quantization noise).
• Clean up legacy Java applet audio — Old educational software, NASA mission archives, and museum interactives that shipped with .au clips can be trimmed without breaking the AudioClip decoder, which only handles a strict subset of AU encodings.
• Trim scientific recordings — Bioacoustic researchers and SETI archives still distribute AU because the format is endian-explicit (always big-endian) and self-describing in a tiny header — useful for FORTRAN/C pipelines that pre-date WAV's RIFF chunks.
• Extract a clip for a NeXTSTEP/OpenStep app — NeXT's Sound.h API expects AU; trimming at the source keeps the file directly usable in vintage software emulators (Previous, OPENSTEP/Mach).
• Keep AU when WAV breaks tooling — Some FTP-era audio servers, ham-radio digital decoders, and academic speech corpora ship parsers that only recognize the .snd magic number. Converting to WAV would require updating the consumer; trimming as AU does not.
• Reduce file size without changing format — A 1-hour 8 kHz μ-law AU is about 28 MB. Trimming to a 30-second highlight with the same encoding drops it to roughly 230 KB — small enough for an email attachment or a Slack upload under the 1 MB free-tier cap.

AU Format at a Glance

Property	Value
Magic number	.snd (0x2E736E64), big-endian
Header size	24 bytes (six 32-bit words) plus optional info chunk
Encodings	8-bit μ-law, 8-bit A-law, linear PCM (8/16/24/32-bit), 32/64-bit IEEE float, ADPCM variants
Typical sample rates	8000 Hz (telephony), 11025, 22050, 44100, 48000 Hz; some NeXT files use 8013 Hz
Channels	Mono (most legacy files) or stereo
Container	None — raw header-prefixed PCM/μ-law samples
Java Sound API support	Yes — read/write via AudioSystem since Java 1.2
File extensions	.au, .snd

AU vs WAV vs FLAC — When to Keep AU

Property	AU	WAV	FLAC
Compression	μ-law / A-law / PCM	PCM / ADPCM	Lossless compression (~50%)
Header	24 bytes, big-endian	44+ bytes RIFF, little-endian	Variable, with metadata blocks
Best for	Telephony, legacy Unix/Java	Modern editing, broadcast	Archival music
Native streaming	Yes — header has known size	Trickier — chunk parsing	Yes
Metadata (tags)	Optional info chunk only	Limited (LIST INFO)	Rich (Vorbis comments)
File size at 8 kHz mono	~28 MB/hr (μ-law)	~28 MB/hr (μ-law in WAV)	~14 MB/hr (PCM compressed)

If your workflow is voice-over-IP, IVR, or any μ-law pipeline, stay in AU. If you need editor compatibility (Audacity, Pro Tools, Logic), convert with AU to WAV after trimming. For long-term archival of music-quality audio, AU to FLAC gives lossless compression and proper tagging.

Quality Preset and Bitrate Reference

Preset / Mode	Typical bitrate	Use case
Quality Preset — Highest	~256–320 kbps equivalent	Music-quality AU at 44.1/48 kHz
Quality Preset — High	~192 kbps	Spoken-word podcasts, lectures
Quality Preset — Medium	~128 kbps	General-purpose voice
Quality Preset — Low / Lowest	~64–96 kbps	Voicemail, IVR prompts
Constant Bitrate — 64 kbps	64 kbps	Telephony-grade μ-law (8 kHz mono)
Constant Bitrate — 32 kbps	32 kbps	Compressed voice, dictation

Frequently Asked Questions

Why is my AU file only 8 kHz mono?

Because that is the historical default. Sun's original /dev/audio device on the SPARCstation 1 emitted exactly 8-bit μ-law at 8000 Hz, and Java 1.0/1.1's audio APIs were locked to that profile. Most AU files older than the early 2000s — voicemail dumps, classroom Java applets, museum exhibits — are still in that shape. Trimming preserves it; if you want higher fidelity, you'll need to re-record from the source rather than upsample.

Should I pick Quality Preset or Constant Bitrate?

Pick Quality Preset for music or general audio — the encoder allocates bits where they matter most and the file size scales with content. Pick Constant Bitrate when a downstream system expects a fixed kbps (telephony bridges, IVR providers, fixed-rate streaming). For 8 kHz μ-law telephony AU, Constant Bitrate at 64 kbps is the canonical setting.

Is mu-law (μ-law) the same as A-law?

They are sibling companding schemes defined in ITU-T G.711, but they use different logarithmic curves. μ-law is the North American and Japanese telephony standard; A-law is the European/ITU standard outside those regions. AU supports both, and the Java Sound API exposes them as AudioFormat.Encoding.ULAW and AudioFormat.Encoding.ALAW. If your file came from a North American carrier, it is almost certainly μ-law; from a European PBX, almost certainly A-law.

Can I trim an AU file without re-encoding the audio?

The xconvert pipeline re-encodes to keep the output bit-exact to your selected Quality Preset / sample rate / channel settings. If you specifically need a stream-copy trim (no re-encoding, instant cut), the FFmpeg command line equivalent is ffmpeg -ss <start> -t <duration> -i input.au -c copy out.au. The trade-off: a stream-copy cut will only land on encoder boundaries, which for μ-law is sample-accurate, but for ADPCM may snap to the nearest block.

Why does my browser refuse to play AU?

No major modern browser ships a built-in AU decoder. Chrome, Firefox, Edge, and Safari rely on system codecs that historically supported AU on macOS (via Core Audio) but never on Windows or Linux without extra software. To preview an AU clip in a browser, convert it first with AU to MP3 or AU to OGG. For local playback, VLC, Audacity, and afplay (macOS) all decode AU natively.

What happens if my file uses an unusual sample rate like 8013 Hz?

Some early NeXT recordings used 8012.821 Hz (rounded to 8013) because the hardware clock divided differently from Sun's exact 8000 Hz. xconvert reads the header rate verbatim and re-samples to whatever you select in Audio Sample Rate. Picking 8000 Hz performs a tiny resample — inaudible for voice. Picking 22050 or 44100 Hz upsamples but does not add new high-frequency content; the original is band-limited to roughly 4 kHz.

Will trimming change the .snd magic number or break legacy parsers?

No. Output retains the standard .snd (0x2E736E64) magic header in big-endian form, with the data offset, size, encoding, sample rate, and channel-count fields rewritten to match the trimmed segment. Any parser that handles standard Sun/NeXT AU — including Java's AudioSystem.getAudioInputStream, libsndfile, and FFmpeg's au demuxer — will read the result.

Can I batch-trim a folder of AU voicemails at the same start/duration?

Yes — drag every file into the upload area and the same Trim window applies to each. This is the typical workflow for cleaning up a directory of .au voicemails (stripping the leading "you have a new message" prompt at a fixed offset). For per-file trim points, run them as separate jobs.

What's the difference between trimming AU and using the generic Audio Cutter?

The trim-AU page is pre-configured for AU output: the format is locked, the encoding options match what AU supports (μ-law, A-law, PCM), and you avoid an accidental WAV/MP3 export. The general Audio Cutter accepts any input but defaults to a different output format. Use this page when the consumer of the clip explicitly requires .au; use the generic cutter when you want flexibility.