Convert AMR to MP3

What AMR is and where it comes from

AMR (Adaptive Multi-Rate) was designed in 1999 for mobile phone calls. It records mono speech at 8 kHz and squeezes it to about 12 kbps, so a minute takes only about 90 KB. That made it the natural choice for:

• voice memo apps on older phones,
• call recording apps,
• dictation and voicemail saved from some phone systems.

Few modern players open AMR. VLC does, but most phones, editors and transcription services want MP3, WAV or M4A instead. Converting to MP3 makes those old recordings playable everywhere again, from a laptop to a car stereo, and easy to send to the people who were on the call. Each MP3 keeps the name of its AMR, so recordings named by date and time stay in order in File Explorer.

What to expect from the MP3

Converting can't add what the recording never had. AMR captures only the range of an ordinary phone call, with nothing above 4 kHz, and the MP3 keeps that same narrow, telephone-like sound. It does not become clearer or fuller.

The MP3 will be about five times bigger than the AMR, since Octoolo writes it at 64 kbps where the AMR used about 12, but it is still small: around 0.5 MB a minute, or about 29 MB for an hour-long call; five hours of recordings from an old phone come to roughly 150 MB. The recording stays mono and keeps its 8 kHz rate unless you pick something else in More settings.

Wideband AMR files, the 16 kHz kind used for HD voice calls, are read as well. Their MP3s keep the 16 kHz rate, and they do sound noticeably better than the narrowband kind, because there was more in them to begin with.

If an MP3 won't play somewhere

An 8 kHz MP3 uses an unofficial extension of the format called MPEG 2.5. Most computers and phones play it without trouble, but a few older car stereos and MP3 players skip such files or refuse them.

Before converting hundreds of recordings for one of those, convert a single file, copy it over and see whether it plays. If it doesn't, convert the batch with Sample rate set to 16 kHz under More settings. The result is a standard MPEG-2 MP3 at the same 64 kbps, and it sounds exactly as before: raising the rate adds no detail, it only makes the file fit the standard. It is also the setting to try if a transcription or upload service turns the 8 kHz file down.

Making old recordings easier to listen to

Octoolo's Audio Editor can help a hard-to-hear recording once it is an MP3. Open it, then use Reduce background noise for hiss and hum, Even out loudness to bring a quiet call up to a normal level, or the volume control, which goes up to twice as loud. When you're writing down what was said, the speed setting from 0.5× to 2× keeps voices at their normal pitch.

Octoolo doesn't turn speech into text itself. Most transcription services accept MP3; if one asks for WAV, pick WAV in the same window instead, and set 16 kHz if it names that rate. Either way, the recording only leaves your PC when you choose to send it somewhere. Keep the AMR files as well: they are tiny, and they are the originals.