Initializing... drag & drop files here
Supports: AAC
For almost everyone, the honest answer is no. AAC is a high-fidelity music codec; AMR is an 8 kHz telephone-quality speech codec, so converting AAC to AMR is a dramatic, audible downgrade — not a way to get "better" or even "smaller, good" audio. Do it only when a specific system demands a .amr file: a voicemail box, an old feature phone, an IVR menu, or embedded telephony hardware. If you just want a smaller AAC file without wrecking it, convert AAC to MP3 or keep AAC instead.
| Property | AAC | AMR (AMR-NB) |
|---|---|---|
| Full name | Advanced Audio Coding | Adaptive Multi-Rate |
| Standardized | ISO/IEC MPEG-2 (1997), MPEG-4 (1999) | 3GPP, October 1999 |
| Designed for | Music and general audio | Telephone-network speech |
| Coding | MPEG transform (MDCT) | ACELP speech model |
| Sample rate | 8–96 kHz | 8 kHz (fixed) |
| Frequency response | Up to ~20 kHz | 200–3,400 Hz |
| Bitrate (typical) | 96–320 kbps | 4.75–12.2 kbps |
| Channels | Up to 48 (stereo, 5.1, etc.) | Mono only |
| Lossy | Yes | Yes (far more aggressive) |
| Best for | Listening, streaming, archiving | Voice memos, MMS, telephony |
| File extension | .aac, .m4a |
.amr, .3ga |
Both are lossy, so AAC to AMR is a second generation of loss: the encoder cannot recover detail AAC already discarded, then throws away everything outside the narrow speech band on top of that.
.amr — voicemail platforms, IVR/auto-attendant prompts, 2G/3G telephony test rigs, and some embedded or two-way-radio firmware accept only AMR..aac files. Batch uploads are supported..amr file. No sign-up, no watermark.Only in raw bytes, and at a steep cost. AMR-NB is tiny because it discards everything above 3,400 Hz and runs at 4.75–12.2 kbps — fine for a voice memo, ruinous for music. AAC already compresses well at 96–128 kbps while keeping full bandwidth and stereo, so if your goal is a smaller file that still sounds good, stay on AAC or use AAC to MP3 rather than dropping to telephone quality.
Yes, substantially, and in two ways. AAC is already lossy, so re-encoding adds a second generation of loss; on top of that AMR narrows the audio to the 200–3,400 Hz speech band at 8 kHz, mono. Speech stays intelligible, but any music, brightness, or stereo image is gone for good. This is a compatibility conversion, not a quality one.
No. AMR is mono by specification, so a stereo AAC track is downmixed to one channel before encoding. AMR is also a bare speech bitstream with no room for cover art, multi-channel layouts, or rich tags. If keeping stereo or metadata matters, convert AAC to M4A or AAC to MP3 instead.
Pick AMR Narrow Band (8 kHz, 4.75–12.2 kbps) for the broadest compatibility with older phones, MMS, and 2G/3G voicemail — it is the default. Pick AMR Wide Band (G.722.2, 6.60–23.85 kbps) when the receiving device supports HD Voice and you want clearer speech; AMR-WB at 12.65 kbps usually sounds noticeably better than AMR-NB at 12.2 kbps. Both are mono and both are built for voice, not music.
That is AMR working as designed. Its ACELP codec models the human vocal tract with very few bits per frame, so it keeps only speech frequencies and, at the lowest modes, can make consonants warble. If the source was music it will always sound hollow through AMR. For spoken-word audio, stepping up to 7.40 or 12.2 KBPS AMR-NB (or an AMR-WB mode) keeps it clean.
In our testing, a one-minute AAC voice clip re-encoded at the 12.2 KBPS AMR-NB mode produced a .amr file of roughly 90 KB with fully intelligible speech. The same pipeline run on a music clip produced an equally tiny file, but the result was thin and hollow — a clear demonstration that AMR is for voice, not songs.
Your file is uploaded over an encrypted connection, converted on our servers, and the upload and result are deleted automatically a few hours after conversion. There is no sign-up, no watermark, and files are never shared or made public.