WAV vs MP3 for Podcast Recording: What to Send Your Editor
What you send your editor sets the ceiling on what comes back. No amount of skill recovers detail that was never recorded, and the most common way detail gets lost is choosing the wrong format before anyone presses record.
The short answer
Record WAV. Send WAV. Deliver MP3.
WAV is uncompressed — every sample of the original recording is preserved. MP3 is lossy: it permanently discards information the encoder judges you will not miss. That judgement is reasonable for listening and unhelpful for editing.
Why it matters more than people expect
MP3 encoding is not reversible. Once the data is gone, every subsequent process works on what remains.
Three things get worse:
- Noise reduction becomes less accurate. The encoder has already introduced small artefacts, and noise reduction treats those as part of the signal. You end up processing the compression rather than the noise.
- Each re-encode compounds. Editing an MP3 and exporting MP3 means encoding twice. Artefacts from the first pass are treated as real audio by the second.
- Sibilants and room tone suffer first. These are the hardest parts for an encoder, and they are exactly where listeners notice damage.
A high-bitrate MP3 from a quiet, well-recorded room is workable. A 64 or 96 kbps file from a noisy room is a rescue job with a hard ceiling.
Settings worth using
| Setting | Recommendation |
|---|---|
| Format | WAV (PCM) |
| Sample rate | 48 kHz for video podcasts, 44.1 kHz for audio-only |
| Bit depth | 24-bit |
| Channels | Mono per microphone |
| Tracks | One file per speaker, unmixed |
Why 24-bit
24-bit gives far more headroom than 16-bit, which means you can record conservatively — peaks around −12 dB — without losing quality when levels are raised later. Recording too hot to “use the full range” is a 16-bit habit that causes clipping, and clipping is unrecoverable.
Why 48 kHz for video
Video runs at 48 kHz. Matching avoids a sample-rate conversion, which is one less process and one less chance of drift over a long recording.
Why one file per speaker
This is the single most useful thing you can do for your editor. With separate tracks, a cough from one guest can be removed without touching anyone else, each voice can be balanced independently, and a noise problem on one microphone stays on that microphone.
With a single mixed file, every fix affects everyone. An editor can still do good work, but the options narrow considerably.
File sizes, and why they are fine
An hour of 24-bit, 48 kHz mono WAV is roughly 500 MB. Two speakers makes that a gigabyte, which sounds alarming and is not a real problem: upload once to any file-sharing service and it transfers in the background.
Compressing to MP3 to save upload time trades a permanent quality loss against a few minutes of waiting. It is never worth it.
If bandwidth is genuinely limited, FLAC is the right compromise — it compresses to roughly half the size and is mathematically lossless, so the editor gets exactly the original samples back.
What to do if you only have MP3
Send it anyway. A good editor works with what exists, and plenty of excellent episodes have been made from MP3 sources.
Two things help:
- Do not re-encode before sending. Send the original file as it came off the recorder, even if it is large. Converting MP3 to WAV before sending adds nothing — the lost data stays lost — but converting MP3 to MP3 makes it worse.
- Say what it is. Knowing the source is a 96 kbps MP3 changes how it should be processed.
Remote interviews
Most remote recording platforms record each participant locally at full quality and upload afterwards, rather than recording the compressed call audio. Check that this is switched on before the interview, not after — it is usually a setting, and the difference is large.
Recording the call itself captures audio that has already been compressed for transmission, with any dropouts baked in.
What to send, in one list
- One WAV per microphone, unmixed
- 24-bit, 44.1 or 48 kHz
- Peaks around −12 dB, never touching zero
- Full unedited takes, including the false starts
- Any music or stings as separate files
- A note about anything you already know is wrong
That last point saves the most time. Telling your editor that a guest’s microphone cut out at 22 minutes is faster than them finding it.
If you want a second opinion on your current setup, send one episode to our podcast editing service and we will tell you what the recording is limiting before you change anything.
