WAV to MP3 converter
Start from the destination, not the codec: who has to listen to this file, and how big may it be? MP3 answers that by discarding samples, and the bitrate is the dial that decides how much listening quality survives the discard.
The target codec is lossy, so the original samples are not restored bit for bit.
Processed on your device. Files aren’t sent anywhere.
Begin with the deliverable and its size limit
Illustrative: Dana records an interview as a 44.1 kHz stereo WAV, and the hosting slot has a fixed allowance. The recording itself is fine; the problem is that a WAV is too large for the slot and for a listener on a phone. So the brief is written first: one MP3, small enough to upload, intelligible in a noisy room. The bitrate is then chosen to satisfy that brief rather than picked out of habit.
Only one setting is exposed here, which is deliberate. Length, sample rate and channel layout come from your file; the encoder decides the rest.
- Drag the WAV onto this page; the local FFmpeg worker loads on first use
- Choose the bitrate from the destination: 96 for speech where size dominates, 128 for podcasts, 192 by default for music, 320 when the file will be edited again
- Convert once — one file per run, up to 60 MiB and 600 seconds of audio
- Listen to the opening and the closing minute before uploading; a smaller file is not by itself a listenable one
Before you compress
Does MP3 keep everything from the WAV?
Which bitrate should I pick?
Will the sample rate change?
Is the recording uploaded to a server?
Why does the MP3 sound hollow at 96 kbps?
Working backwards from the bitrate
Bitrate is a budget expressed in kilobits per second, and total size follows from it multiplied by duration. Choose the number that matches the listener, then accept what the encoder gives up. A podcast host with a modest slot and a phone audience is a different budget from a file that will be normalised later.
| Bitrate | Best when | What you are trading away |
|---|---|---|
| 96 kbps | Voice memos, transcripts, chat attachments where size dominates | High-frequency detail and stereo width |
| 128 kbps | Podcasts and interviews listened to on a phone | Some ambience and quiet passages |
| 192 kbps | Music playback — the default setting | Little that a phone listener notices |
| 320 kbps | A file that will be edited again or played on good equipment | Size — the saving over the WAV shrinks |
What the encoder discards, and what to listen for
A lossy encoder is not compressing the waveform; it is rebuilding a version of it that fits the budget. Quiet passages, reverb tails and the edges of the stereo image are the first things to go, and they go in a way that depends on what is loud at that moment.
- Listen for sibilance and cymbals — these degrade before speech does
- Listen for quiet gaps — they can fill with a watery artefact instead of staying silent
- Nothing is recovered later: a compressed file cannot be turned back into the original samples
Why the length is checked rather than assumed
After encoding, the engine probes the new file and compares its codec and duration with the source. If the duration drifts beyond a small tolerance the run is reported as a failure instead of handing you a truncated MP3, so a bad encode is visible rather than silent.
When MP3 is the wrong deliverable
MP3 is right when the destination is a listener or an upload slot. It is wrong when the file is a master, an edit source, or evidence that has to be measured.
- Archive: keep the WAV, or use WAV to FLAC, which stores the same samples without spending quality
- Editing: give the editor raw PCM — decode an existing MP3 with the MP3 to WAV route
- Analysis: spectrograms and level measurements need uncompressed audio, not a reconstruction