AMR Compressor

How to Compress AMR Audio Online

1. Upload Your AMR Files: Drag and drop or click Add Files to select one or more .amr recordings. Batch is supported.
2. Pick Audio Codec — AMR-NB or AMR-WB: Under Audio Codec, choose AMR Narrow Band (8 kHz sample rate, 200–3400 Hz bandwidth) for the smallest files or AMR Wide Band (16 kHz, 50–6400 Hz) for noticeably clearer speech. Switching an existing AMR-WB recording to AMR-NB roughly halves the file size.
3. Choose Constant Bitrate: AMR uses a fixed set of bitrate modes. For AMR-NB pick from 4.75, 5.15, 5.90, 6.70, 7.40, 7.95, 10.2, or 12.2 kbps — 7.40 kbps is the standardised toll-quality threshold. For AMR-WB choose 6.60, 8.85, 12.65, 14.25, 15.85, 18.25, 19.85, 23.05, or 23.85 kbps. Audio Sample Rate stays at 8000 Hz (NB) or 16000 Hz (WB), and Audio Channel stays Mono — AMR is mono-only by design.
4. Trim (Optional) and Compress: Under Trim, set a start time and duration in seconds or HH:MM:SS.sss to keep just one segment, then click Compress. Files process in your browser session — no sign-up, no watermark.

Why Compress AMR Files?

AMR (Adaptive Multi-Rate) is the speech codec adopted by 3GPP for GSM and UMTS mobile telephony, standardised in 1999 for AMR-NB (3GPP TS 26.090) and 2002 for AMR-WB (3GPP TS 26.190 / ITU-T G.722.2). AMR files are already tiny compared with music codecs, but they are not all equal — a one-minute AMR-WB recording at 23.85 kbps is roughly four times the size of the same minute encoded as AMR-NB at 4.75 kbps. Re-encoding lets you trade clarity for storage in a controlled way.

• Voicemail & call-recording archives — Call-centre and personal voicemail libraries can run into hundreds of gigabytes. Dropping AMR-WB recordings to AMR-NB at 7.40 kbps (toll quality) cuts the per-minute footprint from roughly 175 KB down to about 55 KB while keeping every word intelligible.
• Slow-network voice messaging — At 4.75 kbps an AMR-NB clip needs under 5 kbit/s of bandwidth, so even a 2G EDGE link or a satellite uplink can ship it through. Useful for field reports, remote interview clips, and emergency comms.
• WhatsApp / Signal / Telegram voice notes — Mobile messengers historically saved push-to-talk audio as .amr or Opus-in-.ogg. If you have archived .amr exports from older Android backups, compressing them further (or trimming to the relevant minute) keeps long threads searchable without bloating storage.
• Forensic & legal audio — Court submissions and discovery exports often require the original mobile codec preserved. Re-encoding within AMR (rather than converting to MP3) keeps the file inside the same codec family while you trim or reduce bitrate, which is preferable when chain-of-custody policies forbid format conversion.
• Embedded devices and dictation hardware — Older digital dictation recorders, walkie-talkie apps, and IoT voice loggers natively read AMR. Compressing existing recordings to a lower-tier bitrate before transferring to those devices avoids needing on-device transcoding.

AMR-NB vs AMR-WB — Spec Comparison

AMR Bitrate Modes — File-Size Guide

Approximate per-minute size is the codec rate plus a small container overhead. Use this to pick the lowest tier that still sounds acceptable for your source material.

AMR vs Other Speech Codecs

Frequently Asked Questions

What is the difference between AMR-NB and AMR-WB?

AMR Narrow Band samples at 8 kHz and captures only the 200–3400 Hz telephone band, which is enough for word intelligibility but flattens sibilants and music. AMR Wide Band samples at 16 kHz and processes 50–6400 Hz (with the top 6400–7000 Hz active in 23.85 kbps mode), producing the "HD Voice" sound used on VoLTE. AMR-WB at 12.65 kbps sounds noticeably more natural than AMR-NB at 12.2 kbps and is a better choice when storage is not the priority.

What is the best AMR bitrate for archiving voice recordings?

For AMR-NB, 7.40 kbps is the standardised toll-quality threshold and the most common archival default. For AMR-WB, 12.65 kbps is the typical VoLTE HD Voice mode and a strong balance of clarity and size. Pick lower modes (4.75 or 6.60 kbps) only when bandwidth or storage is severely constrained — speech remains intelligible but artifacting becomes obvious on plosives.

Why can't I select stereo or a higher sample rate?

AMR is defined as a mono speech codec by 3GPP. The encoder operates on a single channel at exactly 8 kHz (NB) or 16 kHz (WB) — those values are baked into the spec rather than being defaults. If you need stereo or music-grade fidelity, convert to a general-purpose codec instead — try AMR to MP3, AMR to AAC, or AMR to Opus.

Will compressing an AMR file twice degrade the audio?

Yes. AMR is a lossy psychoacoustic codec, so each re-encode discards more information. If you must keep the recording inside the AMR family, choose the lowest acceptable bitrate in one step rather than stepping down repeatedly. If you anticipate further edits, convert to a lossless format like AMR to WAV first, edit there, and re-encode to AMR once at the end.

Can I trim an AMR file while compressing?

Yes. The Trim panel takes a start time and duration in seconds or HH:MM:SS.sss. Only the selected segment is kept and re-encoded at the bitrate you choose, so trimming and bitrate reduction stack — a 10-minute AMR-WB call cut to the relevant 2 minutes and re-encoded at AMR-NB 7.40 kbps drops roughly 90% of the original file size.

Why are my WhatsApp voice notes labelled .opus instead of .amr?

WhatsApp switched from AMR-NB to Opus-in-Ogg several years ago — current voice notes save as .opus even though older backups still contain .amr. If your export contains a mix of both, compress the .amr files here and use Opus to AMR (or vice versa) to normalise the collection to one codec before archiving.

Is AMR compatible with iPhones and desktop browsers?

Android devices play AMR natively in the system media stack. iOS, macOS Safari, Windows Media Player, and Chrome on desktop do not play .amr files without a third-party player like VLC. If you need a clip to play in a browser or share with a non-Android user, convert to MP3, M4A, or AAC instead.

Does compressing AMR strip metadata like the recording timestamp?

The audio stream is re-encoded but the AMR container has no standardised metadata fields for timestamps, GPS, or speaker names — most call recorders write that information into the filename or a sidecar .txt/.csv. Keep the original filename if you rely on it for chronological sorting; the compressed output preserves the base name with a suffix.

What is the maximum file size I can upload?

The free tier accepts files up to 200 MB per upload, which comfortably covers multi-hour AMR-WB recordings (roughly 10 MB per hour at 23.85 kbps). For longer single files or batch uploads above 200 MB, an xconvert account raises the limit; see the pricing page for current tiers.