About mp3towave.org

A small site that converts audio without uploading it, and explains the format questions that send people looking for a converter in the first place.

Why this exists

Converting an MP3 to a WAV is a solved problem. Your browser already contains an MP3 decoder, because it needs one to play audio on any web page. Writing a WAV file is a header and a list of numbers. None of that requires a server.

Yet most converters upload your file to a machine you have never heard of, put you in a queue, cap you at two files, hold the result for eight hours, and offer a subscription to make the annoyances stop. For a job the browser can finish locally in under a second, that is a strange amount of infrastructure.

So this site does it in the page. Your file is read from disk, decoded, rewritten, and handed back. It never crosses the network. That is not a privacy feature bolted on afterwards. It is simply what happens when you stop sending the file somewhere it does not need to go.

How it actually works

The details are public because you should be able to check them rather than trust a badge.

• The file is read with the standard File API, the same mechanism any upload form uses, except nothing is posted anywhere afterwards.
• We parse the MP3 frame header for the real sample rate, skipping any ID3 tag at the start. A 48 kHz MP3 then decodes at 48 kHz instead of being silently resampled to whatever your sound card runs at, which is what happens when a converter skips this step.
• Decoding uses the Web Audio API through an OfflineAudioContext created at that sample rate.
• A Web Worker writes the RIFF/WAVE file so the page stays responsive on long recordings, and reports progress as it goes.
• MP3 encoding uses LAME through the lamejs port. LAME has been the reference quality MP3 encoder for two decades and is what Audacity and ffmpeg use. It is LGPL licensed and served unmodified from /vendor/lamejs.min.js, with the licence at /vendor/LAME-LICENSE.txt.
• Batch downloads are packed into a ZIP in your browser, using stored entries with no compression, because audio does not compress usefully a second time.

The WAV writer was checked against ffmpeg rather than trusted by eye. 16-bit, 24-bit, and 32-bit float output all decode as pcm_s16le, pcm_s24le, and pcm_f32le with the right durations, channel counts, and sample rates. The ZIP output was verified by extracting it with Windows and comparing checksums against the originals.

Choices made on purpose

Files convert one at a time. Running ten at once would mean holding ten decoded buffers in memory, which is how tabs crash. Sequential is slower to watch and far more reliable, especially on phones.

16-bit is the default. An MP3 does not hold enough detail to fill 24 bits, so the larger file would be storing zeros. The option is there for when a studio asks.

32-bit float output is left unclamped. Preserving peaks above full scale is the entire reason to choose float. The 16 and 24-bit paths clamp instead, so a hot master cannot wrap around into distortion.

The source sample rate is kept by default. Resampling is arithmetic that can only approximate, so it should be something you ask for rather than something that happens to you.

Constant bitrate for MP3. Variable bitrate gives better quality per megabyte, but constant makes the output size predictable before you start, which matters when you are converting to fit an upload limit.

How the guides are written

Every number on this site is either arithmetic you can repeat or a specification you can look up. File sizes come from the formula, not from a vague sense of how big things are. Commands in the guides were run before they were published.

Where the honest answer is unhelpful, it still gets said. Converting MP3 to WAV does not improve quality, most people cannot pick a 320 kbps MP3 out of a blind test, and 24-bit will not rescue a lossy file. Pages that imply otherwise sell more software, which is presumably why so many of them exist.

Where something is a judgement call rather than a fact, it is written as one. Whether you can hear 192 kbps depends on the material, your equipment, and your ears, and anyone claiming a universal answer is guessing.

How the site is paid for

Running costs are close to nothing, because no server does any conversion work. The site is static files on a CDN. That is the reason there is no file cap and no paid tier: there is no per-conversion cost to recover.

If advertising is added later it will be disclosed here, and it would not change how files are handled. The conversion happens in your browser either way, and there is no version of this site where your audio starts getting uploaded.

Corrections and contact

If something here is wrong, saying so is genuinely useful. Audio format advice is full of confidently repeated errors and this site is not immune to that. Technical corrections with a source will be acted on and the page will say when it was updated.

Bug reports are just as welcome, particularly ones that include the browser, the operating system, and roughly how long the file was. Reach us through the help page or at hello@mp3towave.org.

Changes to the tools and the guides are listed on the changelog.