Initializing... drag & drop files here
Supports: 3GP, 3G2
A .3g2 file is the 3GPP2 container that CDMA2000 handsets — the Verizon and Sprint feature-phone era — recorded to, and its soundtrack is usually not the kind of audio you can just drop into an editor. This tutorial covers pulling that track out and decoding it to WAV: uncompressed linear PCM that opens in any audio application without a codec. It also sets expectations honestly, because what a 3G2 soundtrack can sound like is decided by which vocoder the phone used, not by anything you choose here.
.3gp files are accepted too, and several can be queued to run with the same settings.3G2 and 3GP are the same ISO base media structure split across two standards bodies: 3GPP wrote .3gp for GSM and UMTS phones, 3GPP2 wrote .3g2 in its C.S0050 File Formats for Multimedia Services specification for CDMA2000 phones. The divergence that matters here is audio. On top of the AMR and AAC that .3gp already allowed, C.S0050 permits the CDMA speech codecs: EVRC, EVRC-B, EVRC-WB, 13K (QCELP), SMV and VMR-WB. Which one is inside your file dictates the ceiling on the WAV.
| Audio track in the 3G2 | Typical source | Band | What the WAV can sound like |
|---|---|---|---|
| EVRC, EVRC-B | Voice memos and recorded calls on CDMA handsets | Narrowband, 8 kHz sampling | Telephone-quality speech, clear but with no highs or lows |
| EVRC-WB, VMR-WB | Later CDMA handsets with wideband speech | Wideband, 16 kHz sampling | Noticeably fuller voice; still short of music-grade treble |
| 13K (QCELP), SMV | Older Qualcomm-era recordings | Narrowband, 8 kHz sampling | Telephone-quality speech, often thinner than EVRC |
| AMR-NB / AMR-WB | Clips originating on GSM-side hardware | 8 kHz / 16 kHz | Speech-grade; wideband AMR is the better of the two |
| AAC / HE-AAC | Camcorder-style clips and music captured by an app | Full band | Music-quality PCM, ready to edit |
The EVRC family works on 20-millisecond frames, coding each into 171 bits at full rate — 8.55 kbit/s — and dropping to 80, 40 or 16 bits when the encoder decides speech is quiet or absent. That is an enormous amount of throwing away, and it is done once on the phone, permanently. Converting to WAV decodes those frames into samples and stores them exactly; from that point on nothing further is lost, but nothing lost earlier comes back.
WAV is uncompressed linear PCM, so its size is arithmetic, not a quality setting: sample rate times bit depth times channel count. Our converter writes 16-bit little-endian PCM by default, with 24-bit and 32-bit also selectable under Audio Codec.
| Configuration | Data rate | Roughly per minute |
|---|---|---|
| 8 kHz, 16-bit, mono (a decoded narrowband voice clip) | 128 kbit/s | About 0.96 MB |
| 16 kHz, 16-bit, mono (decoded wideband speech) | 256 kbit/s | About 1.9 MB |
| 44.1 kHz, 16-bit, stereo (CD configuration) | 1,411.2 kbit/s | About 10.1 MB |
| 48 kHz, 24-bit, stereo | 2,304 kbit/s | About 16.5 MB |
The lesson is that upsampling costs real disk space and buys nothing on speech that was captured at 8 kHz. Leaving Audio Sample Rate and Audio Channel on ORIGINAL keeps the file at the bottom row of what it needs to be.
.3g2 files are video-only, and some are the reverse — a voice recording with no picture at all.Carrier-delivered media from the CDMA era was sometimes DRM-wrapped and bound to the handset it was downloaded to; those files will not decode on any tool, and there is no setting that changes that. The other common dead end is truncation — a recording interrupted by a dead battery or a full memory card can leave the file index unwritten, so the container looks valid but has no playable track. Neither is repairable by re-encoding. If your file is a .3gp from a GSM-era phone rather than a CDMA .3g2, 3GP to WAV covers that path, and if you want the picture as well as the sound, convert the whole clip with 3G2 to MP4 instead of extracting audio.
Because the phone encoded it with a CDMA vocoder built for exactly that. EVRC and 13K (QCELP) sample speech at 8 kHz and compress each 20-millisecond frame down to as little as 16 bits, discarding everything outside the telephone band to save network capacity. Decoding to WAV stores those samples with no further loss, but the missing frequencies were never captured. WAV makes the clip editable and universally openable; it does not make it higher fidelity.
16-bit little-endian PCM by default, which is the standard uncompressed configuration and what virtually every editor, DAW and player expects. Under Audio Codec you can move up to 24-bit or 32-bit little-endian PCM if a mastering workflow calls for the extra headroom, though on a decoded narrowband voice clip the extra bits store the same information in more space. Nothing about the choice changes what the vocoder already discarded.
Only when something downstream demands it. Upsampling interpolates the existing audio onto a denser grid — it does not uncover detail that was never sampled, and it multiplies the file size by more than five. If your editor accepts arbitrary rates, leaving Audio Sample Rate on ORIGINAL gives you the same audio in a much smaller file. If it insists on 44.1 or 48 kHz, set it here so you are not fighting a mismatched project rate later.
The decode is, the history is not. The phone's codec threw data away when it recorded, and that is permanent. From the decoded PCM onward, WAV adds nothing and removes nothing — it is an uncompressed store of the samples the decoder produced. So it is accurate to call the WAV a lossless container around audio that was already compressed once, and inaccurate to call the whole chain lossless.
By codec menu. Both containers sit on the ISO base media file format, but 3GPP's .3gp was written for GSM and UMTS handsets and leans on AMR, while 3GPP2's .3g2 was written for CDMA2000 handsets and additionally permits EVRC, EVRC-B, EVRC-WB, 13K (QCELP), SMV and VMR-WB. Both may also carry AAC. The extraction works identically; what differs is which decoder runs and therefore how the result sounds.
Leave Audio Channel on ORIGINAL unless something forces your hand. CDMA speech codecs are mono by construction, so choosing STEREO duplicates one channel into two and doubles the file for no added information. Choosing MONO is worth doing only when the source really is stereo — a music or camcorder-style clip — and you want a smaller file or a single channel on an editing timeline.
The format itself has one: WAV is a RIFF file and RIFF stores chunk sizes in 32 bits, which caps a conventional WAV at roughly 4 GB. In practice you will hit upload size and time over your connection long before that, since it takes hours of CD-configuration stereo to approach the ceiling. If you are only after a section, set Trim to a Time Range instead of converting the whole clip.
Files are uploaded over an encrypted connection, processed on our servers, and deleted automatically a few hours after conversion — no sign-up, no watermark, never shared or made public. If you would rather have a compressed result than an uncompressed one, 3G2 to MP3 produces a much smaller file from the same soundtrack, and the WAV converter hub lists the other formats a finished WAV can go to.