What is an AMR file?

AMR is a codec: the encoding itself rather than the box it travels in. It is used for phones.

The extension is .amr, and the full name is Adaptive Multi-Rate Audio. Both matter less than what the file can hold, which is what the rest of this page is about.

3GPP published it in 1999.

A format that has been readable for that long is a format worth trusting with something you want back in ten years.

It is published in full, so anyone can implement it from the document rather than by inspection, which is why it turns up in so many programs and why files written twenty years ago still open. A published specification is not the same thing as a royalty-free one: where a format wraps a codec, the patent licensing is a separate question the standard does not settle.

It still opens everywhere and is still written by older tools, but nothing new is being built around it. Convert what you mean to keep, and do not choose it for something starting today.

AMR discards detail to stay small, and what it discards does not come back. That is the right trade for something being looked at and the wrong one for something being edited: each re-save takes another bite.

Up to 1. Two covers stereo and is what almost everything is; more than two is surround, and it is the part most likely to be folded down when converting to a format that does not define it.

VLC and FFmpeg read it, and so do most programs of the same kind.

If a file will not open, the format is rarely the problem — it is more often that the program predates it. Converting to something older is the reliable way past that, and it is what the rest of this site is for.

No browser reads it.

That is the single most common reason to convert it: not that the format is bad, but that the place you want to show the file cannot read it.

AMR is meant to be handed over rather than worked in. Editing one is possible and rarely pleasant; the sane approach is to change the source and export again.

The recurring complaints: support outside its home ground is uneven.

None of these is a reason to avoid the format. They are the things worth knowing before you are surprised by one, which is a different claim and a more useful one.

Most audio compression works on the signal: keep what the ear notices, discard the rest. AMR does something different — it assumes the sound is speech and encodes the parameters of a human vocal tract producing it, then reconstructs the voice at the other end from that description.

That is why it achieves rates general audio codecs cannot approach. Speech at 12.2 kilobits per second is intelligible and recognisably the speaker, at roughly a hundredth of the data rate of CD audio. It is also why music through it sounds so strange: there is no vocal tract to model, so the codec describes a guitar as though it were a throat.

Mobile telephony. AMR was standardised by 3GPP in 1999 as the speech codec for GSM and UMTS calls, which means for a decade it was how a very large share of the world’s conversations travelled.

The files people have are the byproducts: voice memos from older Android phones, voicemail exported from a carrier or a PBX, call recordings, and audio pulled out of a 3GP video from a phone of that era. Nothing new records it — modern phones use AAC or Opus for memos — so an AMR file is almost always something retrieved rather than created.

Two variants exist and they sound noticeably different. AMR-NB samples at 8 kHz and covers roughly the telephone frequency range, which is where the muffled quality of an old voicemail comes from. AMR-WB samples at 16 kHz, reaches higher frequencies, and is what "HD Voice" on a mobile network means.

Wideband files usually carry the .awb extension while narrowband uses .amr, though both turn up under either. If a recording sounds like a telephone call, it is narrowband; if it sounds like somebody in the room with a slight compression to it, it is wideband.

Support has quietly disappeared. Browsers do not play AMR, Windows Media Player does not, iTunes and the Apple ecosystem do not, and most modern editors will not import it. VLC will, and so will a handful of other players built around a broad codec library.

The reason is licensing history combined with irrelevance: implementing it meant paying for patents for two decades, and by the time the last of them expired — AMR-NB’s in the second half of the 2010s, AMR-WB’s in 2024 — there was no reason left to add it, since nothing anybody uses produces it. That is why converting is not a preference here but usually the only route to hearing the file at all.

To MP3 or WAV for compatibility, and the result will play everywhere. What it will not do is sound better: the conversion decodes a heavily compressed speech signal and re-encodes it, so the muffled quality, the limited frequency range and any artefacts are preserved exactly. Converting a narrowband AMR to a 320 kbps MP3 produces a large file that sounds like a telephone call.

That is worth setting expectations about, because people frequently convert hoping for an improvement. The realistic goals are compatibility and editability. For speech, a modest bitrate is entirely sufficient — there is no detail above the telephone band to preserve — and a large file buys nothing.

For voicemail, interviews and call recordings, the words are usually the point rather than the sound. Speech recognition handles AMR badly at narrowband rates and handles the converted WAV considerably better, so the useful sequence is convert first, transcribe second.

Expect the accuracy that the source deserves. A wideband recording of one person speaking clearly transcribes well; a narrowband voicemail with background noise and two people talking over each other produces something worth reading as a guide rather than quoting.

Opus, for anything new. It is royalty-free, it handles speech at very low rates as well as AMR does and music properly as well, and it is supported by every browser and every current platform. It is what modern voice applications and messaging services actually use.

AAC where compatibility with older software matters, and WAV when a recording is going to be edited. AMR itself has no remaining use case for new recordings — its value is entirely in the archives that already exist in it.