Initializing... drag & drop files here
Supports: AMR
.amr files come out of old phone voice recorders, voicemail exports, and telephony systems, and they barely open on a modern phone, browser, or editor. AMR is a speech codec — 3GPP adopted it as the standard codec for GSM in the late 1990s — and it was built to carry a voice down a narrow radio channel, not to store audio. AAC is the general-purpose codec that phones, browsers, cars, and desktop players decode natively. Converting makes an old recording usable again; what it cannot do is add back detail the telephony codec never captured.
| Property | Value |
|---|---|
| Full name | Adaptive Multi-Rate, narrowband (AMR-NB) |
| Standardised by | 3GPP, adopted as the GSM standard speech codec in the late 1990s |
| Coding model | ACELP — an algebraic code-excited linear-prediction speech model |
| Sample rate | 8 kHz, 13-bit |
| Audio bandwidth | Band-limited to roughly 200–3,400 Hz — the telephone band |
| Bitrates | Eight modes from 4.75 to 12.2 kbit/s, plus a 1.80 kbit/s silence descriptor |
| Frames | 20 ms frames of 160 samples |
| Channels | Mono |
| File signature | The storage format begins #!AMR\n; the wideband variant begins #!AMR-WB\n |
| Best for | Voice messages and telephony capture, where bitrate matters more than fidelity |
| Property | Value |
|---|---|
| Full name | Advanced Audio Coding |
| Standardised by | ISO/IEC MPEG — MPEG-2 Part 7 in 1997, folded into MPEG-4 Part 3 in 1999 |
| Coding model | MDCT transform coding, designed for music and general audio |
| Sample rates | 8 kHz to 96 kHz |
| Audio bandwidth | Up to the limit of the source, around 20 kHz at CD sample rates |
| Bitrates | Wide range; the Constant Bitrate dropdown on this page runs from 8 kbps to 384 kbps |
| Channels | Mono, stereo, and multichannel |
| Playback | Roughly 97% of tracked browsers, plus iOS, Android, cars, and desktop players |
| Best for | Music and speech that has to play everywhere |
This is a compatibility upgrade, not a fidelity one. An AMR-NB recording was captured at 8 kHz and band-limited to about 200–3,400 Hz at no more than 12.2 kbit/s, so everything above the telephone band was discarded at capture time, before the file ever existed. AAC is a much more capable codec, but it can only re-encode what it is given; nothing in the conversion invents high frequencies, stereo width, or room detail. Plan the settings accordingly — a modest bitrate, mono, and the original sample rate will faithfully carry everything the AMR actually holds, while a high bitrate simply spends bytes describing silence above 3.4 kHz.
.amr files onto the page, or click "+ Add Files" to browse. Batches are supported and all files run with the same settings..aac file, or take the batch as a ZIP. No sign-up, no watermark.No, and no converter can. The audio was captured by a narrowband speech codec at 8 kHz, band-limited to roughly 200–3,400 Hz, at 12.2 kbit/s or less — the detail outside that band was never recorded. AAC can carry what is there faithfully, but it cannot reconstruct what was thrown away at capture. What you gain is playability: an .aac file opens on iPhone, Android, desktop players, and browsers that will not touch .amr at all.
AMR was built for 2G and 3G telephony and voice messaging, not for general playback, so modern iOS, Android, and browser media stacks do not treat it as a standard format — you usually need a dedicated app. AAC is the opposite case: browser-support tracking puts it near 97% globally, with full support in Chrome, Safari, and Edge and partial support in Firefox, which leans on the operating system's decoders. That is why converting is the reliable fix for an old voice memo or voicemail export.
Something low. The source tops out at 12.2 kbit/s for AMR-NB or 23.85 kbit/s for AMR-WB, so there is very little information to preserve: the 20k–32k or 32k–40k Variable Bitrate bands, or a Constant Bitrate of 32–48 kbps, capture essentially everything the original holds. Pushing to 128 kbps or beyond — rates meant for music — produces a much larger file that sounds identical, because the extra bits are describing a frequency range the recording never contained.
Often, yes, and that is worth planning for. We size this from the bitrates rather than guessing: AMR-NB at its 12.2 kbit/s maximum is about 90 KB per minute, while AAC at 32 kbps is about 240 KB per minute — roughly two and a half times larger for audio that sounds the same. A dedicated speech codec will always beat a general-purpose one on speech at these rates. If keeping the file small matters more than compatibility, keep the AMR as well and treat the AAC as the shareable copy.
Keep both on the source values. AMR is single-channel by design, so forcing stereo writes the same signal twice and doubles the size for nothing. The sample rate is the same story: the recording is 8 kHz for AMR-NB or 16 kHz for AMR-WB, and resampling upward to 44.1 kHz does not add bandwidth — it just makes the encoder describe the same narrow band with more samples. Leave Audio Channel and Audio Sample Rate on their original values unless a specific device demands otherwise.
The file announces it. The IETF storage format specifies that a single-channel AMR file starts with the bytes #!AMR\n, while an AMR-WB file starts with #!AMR-WB\n, so opening the file in a text or hex viewer answers the question in a second. Most media players show it too, as an 8 kHz sample rate for narrowband and 16 kHz for wideband. It matters mainly for expectations: AMR-WB reaches roughly 50–7,000 Hz across nine modes from 6.60 to 23.85 kbit/s, so it sounds noticeably fuller than narrowband, though still well short of music quality.
It depends on the destination. AAC gives the broadest native playback on phones and in browsers, which is what most people want for an old voicemail. Choose AMR to M4A if the file is going into a music or voice-memo library, since the MPEG-4 container carries titles and artwork that a raw .aac stream cannot. Choose AMR to MP3 when a platform specifically asks for MP3, and AMR to WAV when feeding an editor or a transcription tool that prefers uncompressed input. For a legacy telephony system that needs the original format back, AAC to AMR runs the reverse.
Files are uploaded over an encrypted connection, converted on our servers, and deleted automatically a few hours after conversion — never shared, never made public, with no sign-up and no watermark. AMR files are tiny by nature, usually well under a megabyte per minute, so uploads are quick even for a long batch of voicemail exports.