Convert CAF to AMR — Free Online Tool
Convert CAF audio files to AMR format using the libopencore_amrnb codec, optimized for speech encoding and mobile telephony. This tool is ideal for extracting voice recordings from Apple's Core Audio Format and compressing them into the narrow-band AMR format used by mobile phones and VoIP systems.
to
FFmpeg Command
Copy this command to run the same conversion locally with FFmpeg on your desktop. Download FFmpeg
Drop your CAF 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
During this conversion, the audio stream inside the CAF container — which may be encoded as PCM, AAC, FLAC, or another codec — is fully decoded and then re-encoded using the libopencore_amrnb codec (AMR Narrow-Band). AMR-NB is a lossy, speech-optimized codec that operates at a fixed sample rate of 8 kHz and is designed for mono voice audio, so any stereo or high-resolution audio in the source CAF file will be downmixed to mono and downsampled significantly. The resulting AMR file is a raw AMR bitstream, a very compact format that strips away the rich container metadata CAF supports. The default bitrate of 12.2 kbps is the highest AMR-NB mode, offering the best quality this codec can deliver while still being extremely small.
What Each Flag Does
| Flag | What it does |
|---|---|
ffmpeg
|
Invokes the FFmpeg multimedia processing engine. In the browser-based version of this tool, FFmpeg runs as a WebAssembly binary (FFmpeg.wasm) directly in your browser with no server involvement. |
-i input.caf
|
Specifies the input file in Core Audio Format. FFmpeg will probe the CAF container to detect the audio codec inside — which may be PCM, AAC, FLAC, or another supported codec — and decode it before re-encoding to AMR. |
-c:a libopencore_amrnb
|
Selects the libopencore_amrnb encoder to produce AMR Narrow-Band audio output. This codec enforces mono audio at 8 kHz, so any stereo or high-sample-rate content from the CAF source is automatically converted to meet AMR-NB's strict constraints. |
-b:a 12200
|
Sets the audio bitrate to 12,200 bits per second, which corresponds to the AMR-NB Mode 7 (12.2 kbps) — the highest quality encoding mode available in the AMR-NB standard. This is specified in raw bits per second rather than the 'k' shorthand because AMR bitrate modes are defined at exact integer values. |
output.amr
|
Defines the output filename with the .amr extension, which tells FFmpeg to write a raw AMR bitstream file. The AMR format has no rich container wrapper, so the result is a compact, mobile-compatible file ready for telephony systems, voice apps, or mobile playback. |
Common Use Cases
- Preparing voice memos recorded on a Mac or iPhone (saved as CAF) for upload to a telephony or IVR system that only accepts AMR files
- Compressing speech recordings captured in high-resolution CAF format for storage or transmission over low-bandwidth mobile networks
- Converting dictation or interview audio from CAF to AMR for use in mobile apps that use AMR as their native voice note format
- Archiving call-center script readings or voice prompts — originally produced in CAF on Apple workstations — into AMR for deployment on mobile phone platforms
- Reducing the file size of spoken-word CAF recordings dramatically for embedding in SMS or MMS messaging systems that support AMR attachments
- Generating AMR audio clips from CAF source files for integration into embedded systems or legacy mobile devices that require AMR-NB input
Frequently Asked Questions
Yes, but the degree depends on your source material and expectations. AMR-NB is a lossy codec engineered specifically for speech at 8 kHz — it discards all audio content above 4 kHz, which is above the range most critical for voice intelligibility. At the default 12.2 kbps mode used here, speech remains clearly understandable, but music, background ambience, or any high-frequency content in the original CAF file will sound muffled or be lost entirely. If your CAF file contains music or broadband audio rather than speech, AMR is not an appropriate target format.
AMR-NB (libopencore_amrnb) is a mono-only codec — it was designed for telephone calls, which transmit a single audio channel. During conversion, FFmpeg automatically downmixes your stereo or multi-channel CAF audio to mono before encoding. For voice recordings this is almost always acceptable since the speech signal is typically identical on both channels. If your CAF file contains stereo music or spatially mixed content, you should consider a different output format rather than AMR.
CAF supports rich metadata such as artist, title, and recording annotations, but AMR is a raw bitstream format with essentially no metadata container infrastructure. All metadata stored in the CAF file — including any markers or region information — will be lost during this conversion. If preserving metadata is important, you should extract and store it separately before converting, or choose a target format like M4A or MP3 that has proper tagging support.
Replace the value after -b:a with one of the valid AMR-NB mode bitrates: 4750, 5150, 5900, 6700, 7400, 7950, 10200, or 12200 (in bits per second). For example, to use the lowest bitrate mode, run: ffmpeg -i input.caf -c:a libopencore_amrnb -b:a 4750 output.amr. Lower bitrates produce smaller files but with noticeably reduced voice clarity, while 12200 (the default used here) is the highest quality AMR-NB mode available.
Yes. On Linux or macOS, you can use a shell loop: for f in *.caf; do ffmpeg -i "$f" -c:a libopencore_amrnb -b:a 12200 "${f%.caf}.amr"; done. On Windows PowerShell, use: Get-ChildItem *.caf | ForEach-Object { ffmpeg -i $_.FullName -c:a libopencore_amrnb -b:a 12200 ($_.BaseName + '.amr') }. Batch processing this way is especially practical when you have large CAF files over 1 GB that exceed the browser tool's limit.
AMR-NB (narrow-band), used in this tool, samples at 8 kHz and is compatible with virtually all mobile devices and telephony systems. AMR-WB (wide-band) samples at 16 kHz, preserving more voice detail and sounding noticeably more natural — it's used in HD Voice calls. If your target system or device supports AMR-WB, you can switch by changing the codec flag to -c:a libopencore_amrwb and using a compatible bitrate such as 23850. However, for maximum compatibility with older phones and legacy systems, AMR-NB is the safer choice.
Technical Notes
CAF was designed by Apple to overcome the 4 GB file size limit of AIFF and WAV, and it supports a wide range of codecs including PCM variants, AAC, FLAC, Opus, and Vorbis. AMR-NB, by contrast, is one of the most constrained audio formats in common use: it is strictly mono, operates at exactly 8 kHz sample rate, and supports only eight discrete bitrate modes ranging from 4.75 kbps to 12.2 kbps. This means the conversion from CAF to AMR always involves significant audio downgrading regardless of the source quality — a 24-bit/96 kHz PCM CAF file and a low-quality 64 kbps AAC CAF file will both converge to the same narrow constraints after encoding. The libopencore_amrnb implementation used by FFmpeg is derived from the 3GPP reference decoder and is well-tested for compatibility with mobile playback environments. Note that AMR files use the .amr extension for single-channel narrow-band content; the format lacks support for subtitles, chapters, or multiple audio tracks, none of which CAF's multi-track capabilities can carry over. File sizes after conversion will be dramatically smaller — a one-minute CAF file at standard PCM quality (~10 MB) will shrink to roughly 90 KB as AMR at 12.2 kbps.