Initializing... drag & drop files here
Supports: WEBA
WEBA is the audio-only stream pulled out of a WebM container, almost always carrying Opus or Vorbis at browser-record-quality bitrates (typically 64-128 kbit/s stereo). It's what Chrome's MediaRecorder API emits when you capture audio in the browser. AMR (Adaptive Multi-Rate), adopted by 3GPP as the standard cellular speech codec in October 1999, is the opposite end of the audio spectrum: a narrowband mono codec engineered to hit toll-quality voice at 12.2 kbit/s and survive at 4.75 kbit/s. Converting WEBA to AMR is almost always about feeding a voice-only system that won't accept anything else.
| Property | WEBA (.weba) | AMR (.amr) |
|---|---|---|
| Container | WebM (Matroska subset) | 3GPP raw / 3GP |
| Typical codec | Opus or Vorbis | AMR-NB or AMR-WB (G.722.2) |
| Sample rate | 8-48 kHz (Opus internally resamples) | 8 kHz (NB) / 16 kHz (WB) |
| Frequency range | Up to ~20 kHz | 200–3400 Hz (NB), 50–6400 Hz (WB; full 50–7000 Hz at 23.85 kbps only) |
| Channels | Mono or stereo | Mono only |
| Bitrate range | ~32-510 kbit/s typical | 4.75-12.2 kbit/s (NB), 6.6-23.85 kbit/s (WB) |
| Best for | Music, podcast, browser recording | Voice, voicemail, telephony |
| Browser playback | Chrome, Firefox, Edge, Safari 14.1+ | None natively — needs decoder |
| Adopted | WebM launched May 2010 | 3GPP standard since Oct 1999 |
| File size (60s speech) | ~700 KB at 96 kbit/s | ~91 KB at 12.2 kbit/s |
| Mode | Bitrate | Variant | Use it when |
|---|---|---|---|
| MR475 | 4.75 kbit/s | AMR-NB | Lowest — barely intelligible, useful for SMS-like voice clips |
| MR59 | 5.90 kbit/s | AMR-NB | Bad cell signal fallback; understandable but compressed |
| MR74 | 7.40 kbit/s | AMR-NB | Toll-quality threshold per 3GPP testing |
| MR122 | 12.2 kbit/s | AMR-NB | Default — full toll-quality narrowband (GSM-EFR equivalent) |
| AMR-WB 12.65 | 12.65 kbit/s | AMR-WB | Wideband entry — clearer "S" and "F" sounds than any NB mode |
| AMR-WB 23.85 | 23.85 kbit/s | AMR-WB | Best AMR quality; only mode that uses the full 50-7000 Hz band |
If you don't know which mode the receiving system needs, pick 12.2 kbit/s AMR-NB. It is the de facto default and what every AMR decoder is required to support.
AMR-NB only encodes the 200-3400 Hz band — the same range as a 1970s landline telephone. Anything above 3.4 kHz (the sparkle in cymbals, the "S" in sibilants, music harmonics) is filtered out before encoding. That is by design: the codec is optimized for intelligibility of human speech at low bitrates, not fidelity. If you need wider frequency response, choose AMR-WB (50–6400 Hz; up to 7000 Hz at the 23.85 kbps mode) by setting the constant bitrate to a wideband mode like 12.65 or 23.85 kbit/s, or convert to a different format such as WEBA to MP3 or WEBA to OGG.
AMR-NB (narrowband, 8 kHz sample rate) is what 99% of legacy systems expect — voicemail, MMS, older Android voice memos, telephony recordings. AMR-WB (wideband, also called G.722.2, 16 kHz sample rate) sounds noticeably clearer for speech but is only supported on newer VoLTE-capable phones and modern softphones. If you're feeding an unknown system, pick NB at 12.2 kbit/s. If you control the playback side and want better quality, pick WB at 12.65 or 23.85 kbit/s.
Not directly in browsers — Chrome, Firefox, Edge, and Safari have no native AMR decoder. VLC, MPV, and Audacity (with FFmpeg) play AMR files. iOS plays AMR through QuickTime; Android plays them natively. If you need browser playback, AMR is the wrong target — convert to WEBA to MP3 or keep the file as WEBA/Opus instead.
For 60 seconds of speech: a typical browser-recorded WEBA at 96 kbit/s Opus stereo is around 720 KB. At AMR-NB 12.2 kbit/s mono it becomes ~91 KB (about 8x smaller). At AMR-NB 4.75 kbit/s it's ~36 KB (20x smaller, but rough on quality). The savings come almost entirely from dropping stereo, narrowing the frequency range, and using a speech-tuned codec — not from a generic compression algorithm.
No. AMR is a CELP-family codec designed around a model of the human vocal tract; it cannot reproduce music acceptably. Drums turn into noise, melodic instruments warble, anything polyphonic falls apart. If your WEBA file is music or a podcast intro with music beds, use Opus, MP3, or AAC instead. AMR is for spoken voice only.
The AMR specification only defines mono channel layouts. Both AMR-NB and AMR-WB downmix any stereo source to a single channel before encoding. The Audio Channel dropdown locks to Mono for that reason. If you need stereo voice (e.g., a two-party call recording with each party on a separate channel), convert to a stereo-capable format like WAV or M4A instead.
AMR uses a fixed set of modes — you can't pick an arbitrary number like 9 kbit/s. AMR-NB has eight discrete modes (4.75, 5.15, 5.90, 6.70, 7.40, 7.95, 10.2, 12.2 kbit/s) and AMR-WB has nine (6.60, 8.85, 12.65, 14.25, 15.85, 18.25, 19.85, 23.05, 23.85 kbit/s). The Constant Bitrate dropdown exposes the valid values. Any other input gets rounded to the nearest mode.
No. Files upload over HTTPS, run through the conversion pipeline, and are deleted on a short timer — typically within a few hours of the download link expiring. We don't index file contents, don't train models on them, and don't require an account to convert.