Convert AU to AMR — Free Online Tool
Convert AU (Sun Audio) files to AMR format using the libopencore_amrnb codec, optimized for speech compression at low bitrates. This is particularly useful for transforming legacy Unix audio recordings into a format compatible with mobile devices and telephony systems.
to
FFmpeg Command
Copy this command to run the same conversion locally with FFmpeg on your desktop. Download FFmpeg
Drop your AU file here
or click to browse
Free — no uploads, no signups. Your files never leave your browser.
Settings
Note: Browser-based encoding uses approximate quality targets. For precise CRF compression, copy the FFmpeg command above and run it on your desktop.
Estimated output:
Conversion Complete!
DownloadHow It Works
AU files typically contain uncompressed PCM audio (most commonly 16-bit big-endian PCM, as defined by Sun Microsystems), which means the source data is raw and lossless. During conversion to AMR, FFmpeg decodes the PCM stream from the AU container and re-encodes it using the libopencore_amrnb codec — an Adaptive Multi-Rate Narrowband encoder optimized for speech frequencies (300–3400 Hz). This is a lossy transcoding step: the AMR codec uses a vocoder-style approach that models human speech patterns rather than preserving the full audio waveform, which is why it excels at voice recordings but degrades music or broadband audio. The output bitrate defaults to 12200 bps (AMR's highest narrowband mode), balancing the best possible AMR quality against its inherent speech-focused compression.
What Each Flag Does
| Flag | What it does |
|---|---|
ffmpeg
|
Invokes the FFmpeg tool — the same underlying engine that powers this browser-based converter via WebAssembly. Running this locally on your desktop gives you identical output to what the tool produces in your browser. |
-i input.au
|
Specifies the input file in Sun AU format. FFmpeg reads the AU header to determine the PCM encoding (e.g., pcm_s16be), sample rate, and channel count before decoding the audio payload for re-encoding. |
-c:a libopencore_amrnb
|
Selects the libopencore AMR Narrowband encoder for the output audio stream. This encoder implements the 3GPP AMR-NB standard — the same vocoder used in GSM mobile voice calls — and is the correct codec for producing standard .amr files compatible with mobile phones. |
-b:a 12200
|
Sets the AMR-NB encoding bitrate to 12200 bits per second, which is the highest of the eight fixed AMR-NB modes. This produces the best quality the narrowband codec can achieve — reducing this value (e.g., to 7950 or 4750) will decrease file size but make speech sound more compressed and robotic. |
output.amr
|
Defines the output filename with the .amr extension, which tells FFmpeg to wrap the encoded AMR-NB bitstream in the standard AMR file container. This file will be directly playable on Android devices, feature phones, and any media player with AMR support. |
Common Use Cases
- Converting historical Unix voicemail or spoken-word recordings stored in AU format into AMR files for playback on mobile phones or feature phones.
- Preparing AU-encoded speech samples or test audio from Unix-based telephony research systems into AMR for submission to mobile network testing tools.
- Archiving old Sun Workstation audio memos or dictation recordings into AMR, significantly reducing file size for storage on constrained devices.
- Converting AU audio from legacy Unix applications into AMR for integration with VoIP systems or IVR (Interactive Voice Response) platforms that accept AMR input.
- Transforming early internet audio clips (commonly distributed as .au files in the 1990s) that contain speech or narration into AMR for use in mobile multimedia messaging (MMS).
- Batch-processing AU recordings from academic speech datasets (a historically common format in linguistics research) into AMR for use in modern speech recognition pipelines.
Frequently Asked Questions
Yes, this conversion is inherently lossy. AU files with PCM audio (the AU default) are uncompressed and bit-perfect, while AMR uses a narrowband vocoder algorithm that discards audio information outside the speech frequency range (~300–3400 Hz). The quality loss is most noticeable with music, sound effects, or high-frequency content — but for clean speech recordings, AMR at 12200 bps (the default used here) delivers intelligible, reasonably natural-sounding voice audio. Once encoded to AMR, the discarded audio data cannot be recovered.
AMR (Adaptive Multi-Rate Narrowband) was designed specifically for mobile telephony, where bandwidth is scarce and the only content being transmitted is human speech. Its codec models the vocal tract rather than encoding the raw waveform, so it efficiently compresses speech but has no mechanism to represent music, wideband audio, or complex soundscapes. If your AU file contains music or mixed audio, the AMR output will sound noticeably thin and distorted — in that case, a format like MP3 or AAC would be a much better target.
AMR Narrowband (libopencore_amrnb) supports exactly eight fixed bitrate modes: 4750, 5150, 5900, 6700, 7400, 7950, 10200, and 12200 bps. The default in this tool is 12200 bps, which is the highest quality AMR-NB mode. To use a lower bitrate — for example to reduce file size for transmission — change the `-b:a 12200` flag in the command to `-b:a 4750` (the lowest mode, roughly half the file size but noticeably more robotic-sounding). For most speech content, 7950 or 10200 bps is a reasonable middle ground.
Yes — AMR Wideband covers frequencies up to 7000 Hz versus Narrowband's 3400 Hz ceiling, producing noticeably more natural speech. To use it, replace `-c:a libopencore_amrnb` with `-c:a libopencore_amrwb` in the FFmpeg command and update the output filename to `output.awb` or keep `output.amr`. Note that AMR-WB requires that your source AU audio was recorded at a higher sample rate (ideally 16000 Hz) — if your AU file was recorded at 8000 Hz (telephony standard), AMR-NB is the appropriate choice.
FFmpeg will handle the sample rate conversion automatically. AMR Narrowband requires a 8000 Hz sample rate, so if your AU file was recorded at 11025, 22050, 44100 Hz, or any other rate, FFmpeg will resample the audio down to 8000 Hz before encoding. This resampling, combined with the narrowband frequency cutoff, means high-quality audio in the AU source will sound significantly lower fidelity in the AMR output — but the conversion will complete successfully without any extra flags.
You can use a shell loop to apply the same command to many files. On Linux or macOS, run: `for f in *.au; do ffmpeg -i "$f" -c:a libopencore_amrnb -b:a 12200 "${f%.au}.amr"; done`. On Windows Command Prompt, use: `for %f in (*.au) do ffmpeg -i "%f" -c:a libopencore_amrnb -b:a 12200 "%~nf.amr"`. This iterates over every AU file in the current directory and produces a matching AMR file — essential when you have a large archive of AU recordings, since this browser tool processes files one at a time.
Technical Notes
AU files use a straightforward header (magic number 0x2E736E64, followed by data offset, size, encoding, sample rate, and channel count) and store audio in big-endian byte order — a reflection of Sun's SPARC architecture. The most common AU codec is pcm_s16be (signed 16-bit big-endian PCM), though the format also supports 8-bit PCM, A-law, and µ-law. All of these are fully decoded by FFmpeg before re-encoding to AMR. The AMR container (.amr) produced by this conversion holds a single mono audio track — AMR-NB is inherently a mono codec, so if your AU source is stereo, FFmpeg will automatically downmix to mono before encoding (left and right channels are averaged). No metadata fields from the AU header (such as annotation strings, which AU optionally supports) are carried over to the AMR output, as the AMR format has no equivalent metadata structure. File size reduction from AU to AMR is dramatic: a 1-minute uncompressed 16-bit mono AU file at 8000 Hz is approximately 960 KB, while the equivalent AMR-NB file at 12200 bps is only about 90 KB — roughly a 10:1 compression ratio.