Initializing... drag & drop files here
Supports: CAF
These two things are not the same kind of thing, which is why the choice confuses people. CAF is Apple's container — a wrapper that can hold almost any audio payload, including uncompressed PCM, Apple Lossless, or even AAC itself. AAC is the codec, the compressed bitstream that phones, browsers, car head units and consoles decode natively. Short answer: keep the CAF as your master if you are still editing in Logic Pro or GarageBand and need its markers and long-recording headroom; convert to AAC the moment the file has to leave the Apple ecosystem or ship at a sane size.
| Property | CAF | AAC |
|---|---|---|
| What it is | Audio container (wrapper) | Lossy audio codec / bitstream |
| Origin | Apple, part of Core Audio on macOS and iOS | MPEG — ISO/IEC 13818-7 (MPEG-2 Part 7, 1997), extended in ISO/IEC 14496-3 (MPEG-4 Part 3) |
| Payload | PCM, Apple Lossless, IMA ADPCM, AAC and more | The AAC bitstream only |
| Compression | Whatever the payload uses — often none at all | Always lossy, perceptual |
| Size ceiling | 64-bit offsets. Apple's spec notes AIFF, AIFF-C and WAV "are limited in size to 4 gigabytes, which might represent as little as 15 minutes of audio", while CAF has no practical limit | Governed by whatever carries it |
| Extra data it can carry | Markers, regions, edit comments, MIDI, channel layout, peak and overview chunks, UMID | Essentially none in a raw .aac stream |
| Typical size, stereo music | ~10 MB per minute at 16-bit/44.1 kHz PCM | ~1.8 MB per minute at 256 kbps |
| Plays outside Apple software | Poorly — most Windows, Android and web players reject .caf |
Natively almost everywhere |
| Best for | Mastering, long captures, Apple-only pipelines | Delivery, sharing, streaming, playback |
.caf onto the page or click "+ Add Files" to browse. Several files can be queued and converted with the same settings in one pass.Sizes below are straight arithmetic from the bitrate, so they hold for any recording of that length.
| Setting | Good for | Size per minute |
|---|---|---|
| 96 kbps | Voice memos, interviews, lecture capture | ~0.7 MB |
| 128 kbps | Podcasts, speech you want to sound polished | ~0.9 MB |
| 192 kbps | Music where size still matters | ~1.4 MB |
| 256 kbps | Music you want to stop thinking about | ~1.8 MB |
| 320-384 kbps | Dense mixes, an intermediate before more editing | ~2.3-2.7 MB |
| Variable Bitrate | Mixed material — spends bits only where the audio is complex | Varies with content |
For most material, yes — that was the design goal of the format. AAC's larger transform window, better stereo coding tools and finer quantization mean a 128 kbps AAC generally sounds closer to the source than a 128 kbps MP3, which is why Apple, YouTube and broadcast platforms standardized on it. The gap narrows as bitrates climb: at 256 kbps and above both formats are hard to distinguish from the source on typical playback gear. Choose CAF to MP3 instead only when the destination is genuinely old.
Some, always — AAC is a lossy codec by definition. What matters is how many lossy generations you have stacked. If the CAF holds uncompressed PCM or Apple Lossless, which is what a Logic or GarageBand bounce normally produces, this is a clean first-generation encode and at 192-256 kbps the difference is very hard to hear. If the CAF already wrapped a lossy payload, you are on generation two, so set the bitrate at or above the original's.
No. The conversion always decodes the payload and re-encodes it, so a CAF that already held AAC is re-encoded rather than repackaged. That means a second lossy generation. If your goal is simply to get that same audio into a file mainstream players accept, CAF to M4A is usually the better target — the M4A container is what music apps expect and it carries tags.
A raw .aac file is the bare bitstream: no tags, no artwork, no chapter marks. .m4a puts the identical AAC audio inside an MP4 container, which is what iTunes, Apple Music, most Android music apps and podcast tooling expect, and it can carry title, artist and cover art. Pick .aac when a tool asks specifically for an elementary AAC stream; pick M4A for anything you plan to keep in a library.
Speech occupies a narrow frequency range, so 96-128 kbps is generally ample and anything higher mostly adds bytes. Setting Audio Channel to mono roughly halves the result again on a single-source recording with no audible penalty. Save 192 kbps and above for music, layered audio, or anything with real high-frequency content.
No. Apple's format can carry markers, regions, edit comments, MIDI data, channel layout, peak and overview chunks and a UMID identifier — a raw AAC stream carries none of that. Keep the original CAF as your master if any of that metadata is load-bearing in your workflow, and treat the AAC purely as the delivery copy.
Our pipeline encodes AAC with the Fraunhofer FDK AAC encoder, which handles high sample rates and up to eight channels, so a 96 kHz or surround CAF does not get forced down to stereo the way some targets require. Bitrate is still worth setting deliberately on a source like that: pick a Constant Bitrate near the top of the dropdown — 256 kbps or above — so a wide-dynamic multichannel recording is not squeezed through a rate meant for speech.
Your CAF is uploaded over an encrypted connection, converted on our servers, and deleted automatically a few hours later. There is no sign-up and no watermark, and files are never shared or made public. The realistic constraint on a multi-hour recording is upload time over your own connection rather than anything about the conversion itself.