MP3 vs WAV
One rule covers almost every case: record WAV if you will edit the file, record MP3 if you will send it. Here is the reasoning, and the exceptions.
What each format is
WAV stores audio as raw numbers — 44,100 or 48,000 measurements per second, each one a 16-bit integer. Nothing is discarded. It is a container for uncompressed data, which is why files are large and why the format has been unchanged since 1991.
MP3 uses a psychoacoustic model to identify parts of the signal a listener is unlikely to perceive, and removes them. The reduction is dramatic — roughly six to one at 128 kbps — and for speech the loss is inaudible. The discarded information cannot be recovered.
| Item | WAV | MP3 (128 kbps) |
|---|---|---|
| Size per minute | 5.3 MB | 0.96 MB |
| One hour | 318 MB | 58 MB |
| Quality loss | None | Some, generally inaudible for speech |
| Editing | Ideal | Degrades on re-export |
| Email attachment | Rarely practical | Fine |
| Encoding time | Instant | A few seconds per minute |
| Compatibility | Universal | Universal |
The rule
- Record WAV when the file is a starting point. Editing, mixing, mastering, transcription of difficult audio, archiving. Every subsequent process works better on a complete signal.
- Record MP3 when the file is a destination. Emailing, uploading, messaging, sharing, or storing a large number of speech recordings.
Why generation loss matters
This is the strongest argument for recording uncompressed. Each MP3 encode discards information permanently. Edit an MP3 and export it again and the second encoder works on already-damaged audio, discarding more. Do it a few times and the degradation becomes clearly audible.
Recording WAV and exporting MP3 once, at the end, means exactly one lossy step no matter how much editing happens in between.
When the difference is actually audible
- Speech at 128 kbps — essentially never. This is why podcasts ship as MP3.
- Music — sometimes, particularly cymbals, reverb tails and quiet passages.
- Singing with sustained notes — more often than people expect. Sustained tones expose encoder artefacts.
- After several encodes — always. Generation loss compounds.
- Sound effects and ambience — frequently, because the encoder's model is built around speech and music.
What about Opus?
Opus is the format your browser produces natively, and it is technically better than MP3 — clearly superior at low bitrates, and free of patents. The reason not to default to it is compatibility: .webm files are rejected by plenty of upload forms and older software. Opus is the right choice for files that stay with you, MP3 for files that travel.
A practical decision
- Podcast recording
- WAV — you will edit it
- Voice note to a friend
- MP3 — small and universal
- Voiceover for a client
- WAV — clients almost always specify it
- Lecture recording
- MP3 — file size across a term matters
- Singing or music
- WAV — artefacts are audible on this material
- Interview for transcription
- MP3 — accuracy is unaffected, size is not
- WAV recorderRecord uncompressed.
- MP3 recorderRecord compressed.
- Podcast recordingWhere format choice matters most.
- Compared with AudacityConverting between the two.
More in guides
- Reducing background noise
- Best microphone settings for voice
- Where browser downloads save
- Microphone permissions explained
See all guides — two kinds of guide: fixing something that is broken, and getting a better result from something that already works.
Questions
Is WAV always better quality than MP3?
Technically yes, since nothing is discarded. Audibly, for speech at 128 kbps, the difference is essentially undetectable. Where WAV genuinely wins is as source material for editing.
Can I convert MP3 back to WAV to restore quality?
No. Converting gives you a large file containing the same degraded audio. The discarded information is gone permanently.
Which is better for transcription?
MP3 is fine. Transcription accuracy depends on clarity, background noise and speaker overlap, not on compression at reasonable bitrates.
Why is my WAV file so large?
Because it is uncompressed — about 5.3 MB per minute of mono audio. That is the format working as intended, not a fault.
What bitrate should an MP3 be?
128 kbps is the standard for spoken word and is what this site produces. Music benefits from 192 or 256 kbps, which is a job for a desktop encoder.
Should I record in Opus instead?
If the file stays with you, Opus is better than MP3 at the same size. If it needs to be accepted by other software or upload forms, MP3 is safer — WebM files are still rejected in plenty of places.