AMR Audio Format: The Mobile Voice Recording Standard Explained

AMR Audio Format: The Voice Recording Standard for Mobile Phones

AMR (Adaptive Multi-Rate) is an audio codec and file format specifically designed for encoding human speech in mobile telecommunications. If you have ever found a .amr file on an old Android phone or received a voice memo from a Nokia handset, you have encountered this codec. It is lightweight, efficient for voice, and deeply embedded in the history of mobile communications.

What Is AMR?

AMR stands for Adaptive Multi-Rate, referring to its ability to switch between multiple bitrate modes dynamically depending on network conditions and channel quality. It is not a general-purpose audio format — it was engineered exclusively for encoding speech signals on mobile networks.

The file extension is .amr for the narrowband variant and sometimes .awb for the wideband variant. The MIME type is audio/amr or audio/amr-wb.

History: The 3GPP Standard for GSM Calls

AMR was standardized by 3GPP (Third Generation Partnership Project) in 1998 and officially adopted as the mandatory voice codec for GSM (2G) and WCDMA (3G) cellular networks. Before AMR, mobile networks used fixed-rate codecs like GSM-FR (Full Rate) and GSM-EFR (Enhanced Full Rate), which could not adapt to varying signal quality.

AMR solved this by defining eight bitrate modes ranging from 4.75 kbps to 12.2 kbps. When a call signal was strong, the codec used higher bitrates for better clarity. When the signal weakened, it dropped to a lower bitrate to maintain connection stability rather than dropping the call entirely. This adaptive behavior was revolutionary for its time.

By 2000, virtually every GSM handset shipped with AMR support. It became the dominant voice codec for 2G and early 3G calls globally, and millions of older Android phones used it as the default format for voice memos recorded with the stock voice recorder application.

AMR-NB vs AMR-WB: Narrowband and Wideband Compared

There are two distinct AMR standards that are often confused:

AMR-WB is also known as G.722.2 in the ITU framework and is the codec behind "HD Voice" on modern carriers. If a call sounds noticeably clearer than a standard phone call, AMR-WB is often the reason.

Why Phones Record in AMR

Older Android devices (Android 2.x through 4.x era) defaulted to AMR-NB for voice memos because:

1. Low storage footprint — A one-minute AMR recording at 12.2 kbps takes roughly 90 KB. The equivalent WAV file at 8 kHz mono takes about 960 KB.
2. Hardware acceleration — The baseband processor already had AMR encode/decode circuits for call processing, so voice recording was essentially free in terms of CPU load.
3. Network sharing — MMS messages have strict size limits; AMR files fit easily within those constraints.
4. 3GPP container compatibility — AMR streams are natively wrapped in 3GP/3GPP containers used in mobile multimedia.

Modern Android phones now default to AAC or Opus for voice memos, but AMR files recorded on older devices remain common.

Why AMR Sounds Poor for Music

AMR is optimized specifically for the human voice frequency range. The codec applies aggressive filtering that removes frequencies below ~200 Hz (cutting bass) and above ~3400 Hz for NB (or ~7000 Hz for WB). Music contains frequencies from 20 Hz to 20 kHz, meaning most musical content is simply discarded.

Additionally, AMR uses linear predictive coding (LPC) which models the human vocal tract. Instrumental sounds, harmonics, and complex timbres that do not match vocal tract resonance patterns are poorly represented. The result is tinny, muffled, and artifact-laden audio when non-speech content is encoded.

Never use AMR for music recording or archival. Use AAC, MP3, FLAC, or Opus instead.

How to Convert AMR to MP3 with FFmpeg

FFmpeg supports both AMR-NB and AMR-WB natively. To convert a single AMR file to MP3:

ffmpeg -i voice_memo.amr -codec:a libmp3lame -q:a 2 output.mp3

For AMR-WB (.awb) files:

ffmpeg -i voice_memo.awb -codec:a libmp3lame -q:a 2 output.mp3

To control the output bitrate explicitly:

ffmpeg -i voice_memo.amr -codec:a libmp3lame -b:a 128k output.mp3

How to Convert AMR to WAV

WAV is the simplest lossless container for raw PCM audio:

ffmpeg -i voice_memo.amr output.wav

This produces a WAV file at 8 kHz mono (matching AMR-NB's sample rate). To upsample to 44.1 kHz stereo for compatibility with some software:

ffmpeg -i voice_memo.amr -ar 44100 -ac 2 output.wav

Note that upsampling does not recover lost frequency content — it only makes the file compatible with software that requires higher sample rates.

Batch Convert AMR Voice Recordings

To convert an entire folder of AMR files to MP3 in one pass:

for f in *.amr; do
    ffmpeg -i "$f" -codec:a libmp3lame -q:a 2 "${f%.amr}.mp3"
done

On Windows (Command Prompt):

for %f in (*.amr) do ffmpeg -i "%f" -codec:a libmp3lame -q:a 2 "%~nf.mp3"

To convert and preserve a log of all processed files:

for f in *.amr; do
    ffmpeg -i "$f" -codec:a libmp3lame -q:a 2 "${f%.amr}.mp3" && echo "OK: $f" >> convert_log.txt
done

Software That Plays AMR Files

Most modern media players support AMR with the appropriate codec:

VLC is the recommended player for AMR files on desktop. It handles both AMR-NB and AMR-WB without any additional codec installation.

Converting AMR on KaijuConverter

KaijuConverter supports AMR-NB input files. Upload your .amr file and select MP3, WAV, OGG, or AAC as the output format. The conversion runs server-side using FFmpeg, so no software installation is needed on your device.

Summary

AMR is a narrow-purpose, speech-optimized codec born from GSM telecommunications. It excels at encoding human voice at very low bitrates, making it ideal for mobile calls and voice memos, but completely unsuitable for music. If you need to work with AMR files, FFmpeg and VLC are the two most capable free tools available.

Related conversions

Audio format pairs that come up most often:

• M4A to MP3
• WAV to MP3
• FLAC to MP3
• AAC to MP3
• OGG to MP3
• MP3 to WAV