An editor asks for a WAV file, but you only have an MP3. Converting it can solve a format-compatibility problem. It does not reconstruct the information discarded when the MP3 was made.
I’m Alex from the Creatune team. Here is a small test using the public Creatune Audio Converter, checked while logged out on October 16, 2026. I used a generated sine wave, not a commercial song. The measurements concern file size and PCM format metadata; I did not perform a listening test or measure perceptual quality.
What changes when you convert?
MP3 is a lossy encoding format. A decoder produces samples from the encoded representation. Saving those samples as PCM WAV stores an uncompressed representation of that decoded audio, including the limitations already present in the source.
A larger file therefore does not establish better sound. Nor does a two-channel file prove that a mono source has gained genuine stereo information.
WAV is a container and can hold different encodings. In this test, the downloaded WAV was 16-bit PCM. Do not assume that every file with a .wav extension has the same properties.
Make an input you can reproduce
This Python 3 script creates an eight-second, 440 Hz sine wave: 48 kHz, one channel, 16-bit PCM.
import math
import struct
import wave
rate = 48000
with wave.open("tone-8s-48k.wav", "wb") as out:
out.setnchannels(1)
out.setsampwidth(2)
out.setframerate(rate)
out.writeframes(b"".join(
struct.pack("<h", round(0.2 * 32767 * math.sin(2 * math.pi * 440 * n / rate)))
for n in range(8 * rate)
))
This WAV is the reference. Keep it. Creating an MP3 here is just a way to obtain a controlled test input; if you already have a lossless master, use that master for a real WAV delivery instead of taking this detour.
Convert in the browser
- Open Audio Converter and choose
tone-8s-48k.wav. - Leave Output format as MP3 and click Download. Save it as
tone-converted.mp3. - Choose that downloaded MP3 as the new input.
- Select WAV and click Download. Save it as
roundtrip.wav.
The page currently lists MP3, WAV, M4A, FLAC, and OGG as inputs, with MP3, WAV, and FLAC outputs. It states that local audio stays on the device. The displayed limit is 50 MB per file and five minutes, or 2.5 minutes on small screens.
The output choice is a format selection. This form did not expose separate sample-rate, bit-depth, or channel controls during the test.
I captured the FLAC selection to show the alternative; the measurements below are for the downloaded WAV, not a tested FLAC export.
Inspect the result, not just the extension
For the two PCM WAV files:
from pathlib import Path
import wave
for name in ("tone-8s-48k.wav", "roundtrip.wav"):
with wave.open(name, "rb") as f:
print(name, {
"bytes": Path(name).stat().st_size,
"frames": f.getnframes(),
"rate": f.getframerate(),
"channels": f.getnchannels(),
"bits": f.getsampwidth() * 8,
"seconds": f.getnframes() / f.getframerate(),
})
These were the actual results of this run:
| File | Bytes | Rate | Channels | Bit depth | Duration |
|---|---|---|---|---|---|
| Original WAV | 768,044 | 48,000 Hz | 1 | 16 | 8.000000 s |
| Intermediate MP3 | 321,984 | Not measured here | Not measured here | Not applicable as a PCM bit depth | Not measured here |
| Round-trip WAV | 1,419,308 | 44,100 Hz | 2 | 16 | 8.045714 s |
The final WAV contained 354,816 frames. Its audio payload size follows the PCM relationship:
354816 frames × 2 channels × 2 bytes = 1419264 bytes
The complete file was 44 bytes larger than that payload. File size is explained by the number of stored samples and their representation; it is not a quality score.
The final duration was about 45.7 ms longer than the original. Encoder delay, padding, and decoding behavior can matter in compressed-audio round trips, but this experiment does not isolate which stage caused this difference. Do not label the files sample-aligned without a separate check. The browser’s rounded “8.0 s” display also cannot reveal a difference this small.
Choose the next step from the actual requirement
- Only WAV is accepted: conversion may be sufficient, followed by a playback check in the destination application.
- 48 kHz mono is required: this measured output does not meet those requirements. Use an export workflow with explicit controls and verify again.
- Better fidelity is required: look for the original lossless source. Rewrapping a lossy source cannot supply it.
- Exact sync is required: compare alignment and duration in the editing timeline; do not infer them from a rounded duration display.
This is one input and one browser run, not a promise that every conversion produces identical parameters. A useful conversion workflow ends with the file requirements you can verify, rather than the size you hoped would imply an improvement.
Top comments (1)
Clear framing, and I like that you separate “format compatibility” from “quality” and report only what you measured. A few ideas that could tighten the test:
Find which stage changed the rate and channels. The round trip went from 48 kHz mono to 44.1 kHz stereo, but the MP3 stage wasn’t measured. Running ffprobe tone-converted.mp3 would show whether the resampling and mono-to-stereo change happened during MP3 encoding or during WAV export.
Bitrate is a hidden variable. By my arithmetic from the reported size (321,984 bytes) and duration, the MP3 is roughly 320 kbps. That’s a high setting, so audible loss would be small. Mentioning the bitrate, or checking whether the tool exposes it, would help readers compare.
The 45.7 ms of extra duration is probably codec delay and padding. MP3 encoders such as LAME store encoder delay and padding in a header tag, and whether the decoder trims it determines the final length. Comparing the decoded length with and without that metadata would isolate it.
A sine wave hides lossy artifacts. A single tone is easy for a perceptual codec. A log sweep or noise burst plus a spectrogram would show the low-pass cutoff and other changes, so you could show quality loss and not only metadata.
The 44-byte difference being the canonical WAV header is a nice sanity check. Would you add a spectrogram comparison in a follow-up?