Convert audio files
Convert between MP3, WAV, OGG, M4A and FLAC with bitrate, sample-rate and channel control plus optional loudness normalisation. Fully local.
- Runs in your browser
- Files are never uploaded
- Free, no account
First use downloads the 10 MB video engine from this site. After that it is cached and works offline. It is WebAssembly served from onthisdevice.com — no third-party CDN — and your file is still processed entirely on your own device.
How to convert an audio file
- Drop in an MP3, WAV, OGG, M4A, FLAC, AAC, Opus or WMA file — or a video, if you only want its soundtrack.
- Choose the output format. The estimated size updates immediately so you can see the trade-off.
- Set a bitrate for lossy formats, a bit depth for WAV, or a compression level for FLAC.
- Open Advanced options if you need to change the sample rate, force mono, or normalise loudness.
- Convert, listen to the result in the page, and download it.
Format reference
| Format | Codec | Lossy? | Size for 3 minutes | Choose it when |
|---|---|---|---|---|
| MP3 | LAME | Yes | 4.3 MB at 192 kbps | Anything has to be able to play it |
| M4A | AAC | Yes | 4.3 MB at 192 kbps | Better quality than MP3 at the same bitrate |
| OGG | Vorbis | Yes | 4.3 MB at 192 kbps | Patent-free pipelines, games, Linux tooling |
| FLAC | FLAC | No | ≈ 18 MB | Archiving; roughly half the size of WAV |
| WAV | PCM | No | 31 MB at 16-bit/44.1 kHz | A tool demands raw PCM input |
The one rule of lossy conversion
Every lossy encode throws away data that cannot come back. Converting MP3 to FLAC does not restore anything — it stores the already-damaged signal losslessly, in a much bigger file. Converting MP3 to MP3 at a higher bitrate is worse than useless: you pay for the size and take a second round of damage.
So: keep your masters in FLAC or WAV, and generate lossy copies from them. When you only have a lossy file and something demands another lossy format, convert once, directly, at a bitrate at least as high as the original.
Sample rate and channels
Sample rate sets the highest frequency that can be represented — 44.1 kHz covers human hearing up to about 22 kHz, 48 kHz is the video and broadcast standard, and 16 kHz is enough for intelligible speech. Downsampling a 48 kHz file to 44.1 kHz is harmless; upsampling adds nothing but bytes.
Channels: forcing mono halves the data for a genuinely single-source recording such as a lapel microphone or a phone voice memo, with no perceptible loss. Do not do it to music.
Loudness normalisation
The optional normalisation pass applies FFmpeg's loudnorm filter targeting
−16 LUFS integrated with a −1.5 dBTP ceiling — the level most podcast platforms expect, and a
reliable fix for interviews where one speaker is far quieter than the other. It analyses and
corrects in a single pass, so it is not quite as precise as a two-pass measurement, but it is
well within the tolerance those platforms apply. Leave it off for music that has already been
mastered.
Questions
Which format should I convert to?
MP3 for maximum compatibility, M4A (AAC) for better quality at the same bitrate, FLAC when you need a lossless archive at about half the size of WAV, and WAV when a program demands raw PCM.
Does converting FLAC to MP3 lose quality?
Yes — MP3 is lossy, so detail is discarded permanently. Converting the other way (MP3 to FLAC) does not restore anything; it just wraps the already-degraded audio in a bigger file.
What does loudness normalisation do?
It applies the EBU R128 loudnorm filter targeting −16 LUFS, the level most podcast platforms expect. It is ideal for voice recordings and best left off for music that has already been mastered.
Can I feed it a video file?
Yes. Drop an MP4, MOV or MKV and the video track is ignored — only the soundtrack is decoded and re-encoded into the format you picked.