The five output formats and what each one is
MP3 is encoded with LAME at 192 kbps constant bitrate. It is lossy, it is nearly forty years old in concept, and it remains the format with the broadest device support, which is why it is still the safe default for anything that has to play somewhere unknown.
M4A here means AAC audio in an MP4 container, encoded at 192 kbps. AAC is the successor to MP3 and is more efficient at the same bitrate, so a 192 kbps AAC file generally sounds better than a 192 kbps MP3. It is the native format of the Apple ecosystem and is supported almost everywhere modern.
OGG means Vorbis at 192 kbps, a royalty free lossy codec common in games and on open platforms. WAV is uncompressed 16 bit PCM: no encoding decisions, no loss, and very large files. FLAC is lossless compression, typically storing the same samples as a WAV in 50 to 60 percent of the space, and it is the right choice for archiving.
Lossy, lossless and generation loss
The distinction that matters most is whether the conversion can throw information away. WAV and FLAC are lossless, so converting a WAV to FLAC and back gives you identical samples. The lossy formats work by modelling what the ear will not notice and discarding the rest, and that discarded detail cannot be recovered.
The practical consequence is generation loss. Converting an MP3 to another MP3, or an MP3 to AAC, decodes the already-degraded audio and then applies a second round of psychoacoustic discarding. The artifacts of the first encode are treated as signal by the second. One conversion is usually inaudible on speech and acceptable on music, but repeated round trips accumulate damage.
Converting a lossy file to WAV or FLAC is also worth understanding. It does not restore anything. It produces a large file that losslessly preserves the already-lossy audio, which is useful when a tool downstream requires an uncompressed input, and pointless as an archival step.
A worked example: a 40 minute recording
Take a 40 minute interview recorded as a 48 kHz 16 bit stereo WAV. Uncompressed, that is roughly 440 MB. Converted to FLAC it might land near 250 MB with every sample preserved. Converted to MP3 at 192 kbps it becomes about 55 MB, and to M4A at the same 192 kbps a similar size with somewhat better fidelity.
For spoken word, 192 kbps is generous, and the choice between MP3 and M4A comes down to where the file has to play rather than to audible quality. For music that will be listened to critically, the lossless path is the one to keep, with a lossy copy generated separately for portable use.
The bitrate is fixed at 192 kbps for the three lossy formats. There is no bitrate selector, no variable bitrate option, no sample rate or channel conversion, and no normalization or trimming. This is a format converter and nothing more.
Input handling and the engine
The file picker accepts audio files generally, and the conversion runs through ffmpeg, so the range of inputs it can decode is broad and includes formats not offered as outputs. The audio track of a video file will typically decode as well, which is a practical way to pull the audio out of a recording, though the tool is presented and named as an audio converter.
The ffmpeg engine is compiled to WebAssembly and served from this site rather than a third party CDN. It is fetched on your first conversion, which is a noticeable wait on a slow connection, and served from the browser cache afterwards. Your audio file itself never leaves the device.
One file is converted per run, with no batch queue, and the engine is terminated after each run so a failed conversion does not leave a worker and its memory behind. Long files take longer because the engine runs single threaded in the tab, and the tab has to stay open for the duration.
Which format to pick
Pick MP3 when compatibility is the only consideration: old car stereos, cheap players, upload forms that name the format explicitly. Pick M4A when the destination is a phone, a modern browser or an Apple device and you want better quality per byte at the same bitrate. Pick OGG when you are targeting a platform that prefers Vorbis, such as some game engines and open source stacks.
Pick WAV when a downstream tool demands uncompressed PCM, for example an older editor, a transcription pipeline or an embedded device, and accept the file size. Pick FLAC when you are storing a master copy, since it is lossless, widely supported by media software, and roughly half the size of WAV.
Once you have converted a lossy file, keep the original if there is any chance of needing another format later. Converting again from the original avoids stacking a second generation of loss on top of the first.