MP3 vs WAV vs FLAC: Choosing an Audio Format That Fits the Job

WAV is not really a compression format. It is a container that usually holds uncompressed PCM samples, which is why a WAV is large: every sample of every channel is stored at full precision. It is the working format of recording, editing and interchange, because every tool reads it and nothing is lost. FLAC compresses those same samples losslessly, typically halving the size, and decodes back to bit-identical audio. It is the right archive format when you want to keep everything without keeping a huge file.

MP3, AAC, OGG Vorbis and Opus are lossy. They discard information the ear is unlikely to miss, using psychoacoustic models, and the result is typically a tenth of the WAV size. The detail is gone for good; decoding an MP3 to WAV produces a large file that contains exactly the same limited information. That asymmetry is the single most important fact about audio conversion.

At 128 kbps, an MP3 is acceptable for speech and background listening and audibly imperfect on cymbals, strings and complex mixes. At 192 kbps most listeners cannot tell it from the source on ordinary equipment. At 256 to 320 kbps the difference is inaudible to nearly everyone, and the file is still a fraction of WAV. The curve flattens: going from 128 to 192 buys a lot, 192 to 320 buys very little. For spoken word, a mono file at 64 to 96 kbps is entirely adequate and halves the size again.

Sample rate matters differently. 44.1 kHz covers the full range of human hearing; 48 kHz is the video and broadcast standard; higher rates exist for production headroom, not for delivery. Downsampling 96 kHz to 44.1 kHz is a real conversion that needs proper filtering, and it is the one place where a careless converter can introduce audible artefacts. Bit depth is about dynamic range: 16-bit is CD standard, 24-bit is for recording and editing, and converting 24-bit to 16-bit requires dithering to avoid quantisation distortion in quiet passages.

For a podcast, MP3 at 96 to 128 kbps mono or 192 kbps stereo is universally playable and small. For music distribution, the platform will specify its preferred format, often WAV or FLAC for masters and AAC for streaming. For a website, MP3 remains the safest single source, though Opus and AAC give better quality at the same bitrate where the browser supports them. For video editing, keep WAV or FLAC until the final export, because every lossy generation accumulates.

Phone and voice recordings add a wrinkle: they are often mono, low sample rate and already compressed. Converting one to a high-bitrate stereo MP3 cannot restore anything, and it triples the size. Check the source properties before choosing output settings, and match them rather than exceeding them.

Do not convert MP3 to FLAC expecting better sound, and do not convert it to WAV expecting a master. Both produce larger files containing the same lossy audio, and the FLAC or WAV label then misleads whoever receives it into thinking they have an original. The only way to regain quality is to go back to a lossless source, if one exists.

Do not chain lossy conversions either. MP3 to AAC, or AAC to MP3, applies the model twice and the second pass spends its bits encoding the artefacts of the first. Convert once, from the best source you have, to the format the destination needs. If you must change container without re-encoding, that is a remux, and it preserves the audio exactly — but it only works when the codec is already supported by the target container.