Audio converter

WAV and FLAC are both lossless: FLAC is simply WAV packed better, typically to around half the size, with every sample recoverable exactly. Converting between them changes nothing you could ever measure, and is the right move for archiving or for handing audio to someone who will edit it.

MP3, AAC and Opus are lossy — they discard detail deliberately to be small, and that detail does not come back. Converting an MP3 to FLAC produces a large file containing exactly the same audio the MP3 had, losses included; it does not restore anything. If quality matters, convert from the original rather than from a copy that has already been through a lossy pass.

Audio files carry metadata — title, artist, album art, and often the software that wrote them or an identifier from where they were bought. Conversions here carry across what makes sense and drop what does not apply to the target format.

If removing that information is the actual goal rather than a side effect, the metadata remover is the tool for it: it shows you what a file is carrying before it takes anything out, and it rewrites the tags without re-encoding the audio, so the recording itself is bit-for-bit unchanged.

Choosing a higher bitrate than the file already has produces a bigger file containing exactly the audio that was there before. The detail a lossy encoder discarded is gone from the file; nothing downstream can consult the original recording to put it back.

This matters most when a file has been through two lossy passes. Each one works on what the last one left, so the artefacts compound. If you have any access to the original recording, converting from that beats converting from a copy, however generous the settings on the second pass.

Nearly all music is recorded at 44,100 samples a second, which is the compact disc rate and was chosen in 1980 because it is comfortably more than twice the highest frequency most adults hear. Video work uses 48,000 instead, which is why audio pulled out of a film usually arrives at that rate.

Converting between the two is resampling and it is not free — the samples have to be recalculated rather than copied. It is inaudible in practice, but it is a reason to leave the rate alone unless something downstream insists on a particular one. Most editors and players accept both without comment.

A stereo file carries two independent streams and costs roughly twice what the same material costs in mono. For music that is the whole point. For a lecture recording, a voice memo or an interview captured on a single microphone, the second channel is usually a copy of the first and is paying for nothing.

Dropping a spoken-word recording to mono halves the file with no audible loss, which on an hour of audio is a difference worth having on a phone. The check is quick: if the recording came from one microphone, there was never stereo information in it to lose.