Convert MP3 to WAV

A lot of people arrive here expecting an improvement, and the honest answer is that there is not one available. MP3 compression works by permanently throwing away parts of the recording that the ear is least likely to miss. That information is gone from the file — not hidden, not compressed, gone. A WAV made from an MP3 contains precisely what the MP3 contained, written out sample by sample without further compression.

The result is a file around ten times larger that sounds identical, because it is identical. This is not a limitation of any particular converter; it is what "lossy" means. The only way to get the missing detail back is to obtain the recording again from a source that still has it.

The first is compatibility, and it is the common one. Audio editors, digital audio workstations, game engines, broadcast systems, telephone platforms and a great deal of older hardware want uncompressed PCM and will simply refuse an MP3. Audacity, Pro Tools, Unity and countless others are happier with WAV, and converting is the whole answer.

The second is editing. Every save as MP3 compresses again, and the damage accumulates — an audio file that has been through five edit-and-export cycles as MP3 is audibly worse than one that has been through one. Converting to WAV first, working there, and exporting to MP3 once at the end means you compress exactly once. That does not improve the source, but it stops it getting worse.

The WAV keeps the sample rate of the original — 44,100 Hz for almost all music, sometimes 48,000 for audio taken from video — at 16 bits per sample, in stereo if the source was stereo. That combination is CD quality and is what virtually every program expects to receive.

Raising either number is possible in principle and pointless in practice. Resampling to 96 kHz or padding to 24 bits makes the file larger and adds no information, because the information is not there to add. If a piece of software demands a specific rate, it will usually resample on import itself, and it will do it with more context than a converter has.

Uncompressed stereo audio at CD quality is close to ten megabytes per minute. A four-minute song is about forty megabytes; an hour-long interview is over half a gigabyte. There is no quality setting to adjust, because WAV does not compress at all — that is the entire distinction between it and every other audio format on this site.

If the size is a problem but you still need something lossless, FLAC is the answer: it compresses without discarding anything, typically to about half the size of the equivalent WAV, and every audio editor made in the last fifteen years reads it. The only reason to prefer WAV is software that insists on it.

Your browser already contains an MP3 decoder — it uses one every time a web page plays audio. This page uses that decoder and then writes the samples out as a WAV, which is a container so simple that no library is needed for it. Neither step requires a server.

So the file is never uploaded, there is no queue, no account and no daily limit. That matters most for the batch case: converting a folder of forty recordings through a site that uploads them means forty uploads and a cap. Here it means leaving the tab open for a moment.

This is a third situation, and a newer one. Whisper builds, voice-cloning services, forced-alignment tools and a good deal of speech research tooling accept WAV and refuse MP3 — not because the audio is better, but because reading uncompressed samples requires no decoder and removes a class of bugs from their pipeline. The requirement is about their code, not your recording.

The rule from the top of this page still applies unchanged: the WAV will be about ten times the size and will contain exactly the audio the MP3 contained. Nothing is recovered. That is fine here, because these tools are reading the speech rather than judging its fidelity, and a converted MP3 transcribes as well as it ever did.

MP3 encoding pads the beginning and the end of the audio. The encoder works in blocks, a recording rarely ends on a block boundary, and so a short run of silence is added to fill it out — typically a few dozen milliseconds at each end.

For speech and most music this is invisible. It matters in exactly two places: a track meant to loop seamlessly, and a set of tracks meant to run into one another without a gap. If either is your case, trim the silence in an editor afterwards, because the WAV preserves the padding as faithfully as it preserves everything else.

The WAV keeps the channel layout of the MP3. A mono recording stays one channel and a stereo one stays two, which is what any receiving program expects.

Most music MP3s are encoded as joint stereo, a technique that stores what the two channels have in common once instead of twice. That is undone during decoding, so what comes out is ordinary two-channel stereo. Nothing about it needs handling — it is simply worth knowing that joint stereo is not a third kind of file.

An MP3 carries ID3 tags, and a great many MP3s carry embedded artwork as well. WAV has no equivalent that anything reliably reads, so the title, the artist and the picture do not come across.

That is a real loss if you are converting a music library and a non-issue if you are converting something in order to edit it. If the tags matter, FLAC is the lossless format that keeps them, and every editor made this century reads it.

Drop up to a hundred files at a time. Each becomes its own WAV and they come back as a ZIP. Nothing is uploaded, so there is no queue behind anybody else and no daily allowance to spend.

Watch the total, though. Ten megabytes a minute adds up faster than people expect: a folder of forty three-minute tracks is well over a gigabyte of WAV being assembled inside your browser. In that range, convert in batches.