Audio Converter
Decode anything the browser can read and write it back as WAV or MP3, with an honest table of what a browser genuinely cannot encode.
Drop audio files here
MP3, M4A, OGG, FLAC, WAV or WebM. Decoded in this page, never uploaded.
MP3 allows only 32, 44.1 and 48kHz.
Scales everything by one factor so the loudest sample sits at 0.98.
- Files
- —
- Total length
- —
- Source size
- —
- Output size
- —
| Format | Read | Write |
|---|---|---|
| WAV | Yes | Yes Written here, byte by byte. Lossless and instant. |
| MP3 | Yes | Yes Written by a 150KB encoder loaded on first use. |
| M4A / AAC | Yes | No No browser can encode AAC. Decode only. |
| OGG / Opus | Yes | No Encodable only in real time, which is slower than useless. |
| FLAC | Yes | No Decoded by the browser; no encoder exists in one. |
| WebM | Yes | No Recording format only. See the Audio Recorder. |
Everything here runs on your device. Nothing you enter is uploaded or stored.
Audio conversion in a browser has a shape nobody explains, and the shape is the reason most online converters make you upload. A browser can read almost every audio format ever shipped. It can write two. Everything else you see offered is either a server doing the work with your file, or a page that fails at the last step after taking your time.
This tool does the part that genuinely works locally, and prints the rest as a table.
What comes in
MP3, M4A, AAC, OGG, Opus, FLAC, WAV and WebM, all decoded by the browser’s own audio pipeline. Drop in a folder if you want; the queue holds as many as your memory allows, each listed with its duration, its resolved output settings and an estimated size.
What goes out
WAV, written here byte by byte — a 44-byte RIFF header and the samples exactly as they were decoded, at 16, 24 or 32-bit. It is instant, it is lossless, and it is large.
MP3, through a 150KB encoder that downloads the first time you ask for it and is then cached. Bitrates from 128 to 320kbps. Because MP3 means compression, the page warns you when your source was already compressed, since that stacks a second generation of loss on top of the first.
The formats that cannot be written
| Format | Read | Write | Why |
|---|---|---|---|
| WAV | Yes | Yes | Written here, byte by byte |
| MP3 | Yes | Yes | 150KB encoder, loaded on first use |
| M4A / AAC | Yes | No | No browser can encode AAC |
| OGG / Opus | Yes | No | Real-time only, which is slower than useless |
| FLAC | Yes | No | Decoded by the browser; no encoder exists in one |
| WebM | Yes | No | Recording format — see Audio Recorder |
Those four are absent from the output menu rather than present and broken. If you need an M4A, the honest answer is a desktop application, and saying so is more useful than a spinner that never finishes.
Rate and channels are separate decisions
Format, sample rate and channel count are chosen independently, because they solve different problems. A voice recording archived as 16kHz mono WAV is roughly a twelfth the size of a 44.1kHz stereo original and loses nothing that carries speech. A music file dropped to mono saves half the bytes and, on a phone speaker, half of nothing.
Folding to mono averages the channels rather than summing them, which is the safe convention — summing two channels can exceed full scale even when neither channel did. The cost of averaging is level: identical channels lose nothing, a hard-panned mix loses up to 6dB. The page tells you the figure for your file, and normalising puts it back.
Estimates before, not after
Every row shows what that file will weigh before anything is encoded. For WAV it is arithmetic and exact. For MP3 it is bitrate times duration, and nothing else — channels and sample rate do not affect it at all, because a constant-bitrate encoder spends the same bits per second regardless of what it is spending them on. Stereo at 192kbps and mono at 192kbps produce the same size; the stereo file just has less to spend per channel.
Nothing is uploaded
Decoding uses the browser’s Web Audio API, the samples are transformed in the page, and the encoder runs on your machine. No server sees the file, which matters more for audio than for most things — audio is often a recording of a room with people in it who did not agree to a third party holding a copy.
For trimming before converting, MP3 Cutter works on the same decoded samples. For joining several files into one, use Audio Merger.
Questions
Why can it open an M4A but not save one?
Because a browser can decode far more formats than it can encode, and almost nothing on the web says so. Decoding is handled by the browser's own audio pipeline, which reads MP3, AAC, M4A, OGG, Opus, FLAC, WAV and WebM. Encoding has no equivalent. The only encoder built into a browser is MediaRecorder, which works in real time — writing ten minutes of audio takes ten minutes — so it is useless for conversion. That leaves WAV, which is written here byte by byte, and MP3 through a small encoder library. AAC cannot be produced by any browser at all. The support table on this page states all of it, because a converter that offers a format it cannot deliver has wasted your time twice.
Does converting an MP3 to WAV improve it?
No, and this is the most common misunderstanding in audio. MP3 compression works by discarding information it judges inaudible, permanently. Converting to WAV writes the surviving samples into a larger container without recovering anything — a 3MB MP3 becomes a 30MB WAV that sounds exactly the same. WAV output is still the right choice in two cases: when the file is going into an editor for more work, so no further compression generations accumulate, and when something in your chain demands WAV. It is never a quality upgrade, and the page says so when you choose it.
What does normalise actually change?
It finds the loudest single sample in the file and scales everything so that sample sits just under full scale, at 0.98. Every other sample moves by the same factor, so nothing about the balance or dynamics changes — it is a volume control that has worked out its own setting. It is useful for a quiet recording, and worth pairing with a fold to mono, because that fold averages the channels and costs level whenever they differ. The page reports how much level the fold will cost in decibels before you convert.
Why does the MP3 sample rate sometimes change on its own?
Because MPEG-1 Layer III defines exactly three legal sample rates — 32000, 44100 and 48000 Hz — and nothing else can be written honestly. Ask for a 22050Hz MP3 and any encoder must either refuse or write a header claiming a rate the data does not have, which plays at the wrong speed in some players and not at all in others. This tool moves the rate to the nearest legal one and tells you it did. WAV has no such constraint, so a 16kHz or 8kHz WAV is perfectly valid and is the right output for a voice archive.
How many files can it do at once?
As many as your device's memory allows, and the constraint is decoding rather than encoding. Every file is decoded to 32-bit floating point samples, roughly 10MB per minute of stereo at 44.1kHz no matter how small the source was, and they are all held at once so the queue can be reordered and re-estimated. Ten three-minute songs is around 300MB, which a desktop handles comfortably. Files are encoded one at a time and packaged into a single ZIP, because a browser treats a burst of downloads as popup behaviour and blocks everything after the first.
Last updated