Lossless vs lossy audio
Audio formats split into two groups, and the split matters more than any difference within either group. Some formats keep every bit of the original recording. Others permanently discard parts of it to save space. Knowing which is which tells you what to keep, what to send, and which conversions you can undo.
The two groups
Lossless — nothing is discarded: WAV, FLAC, ALAC, AIFF
Lossy — information is thrown away permanently: MP3, AAC (M4A), Vorbis and Opus (OGG), WMA
Lossless formats can reconstruct the original recording sample for sample. Lossy formats cannot, and no amount of converting will change that.
What "lossy" actually throws away
Lossy encoders are not crude. They are built on psychoacoustics — research into what human hearing can and cannot detect. If a quiet sound occurs immediately after a loud one at a similar frequency, you will not hear it, so it need not be stored. Frequencies above the range of most adult hearing can go too.
Done well, the result is remarkable: an MP3 at a third of the size that most people cannot distinguish from the original. But the discarded information is gone from the file. Not compressed, not hidden — absent.
Lossless still compresses — it just does it honestly
A common misunderstanding is that lossless means uncompressed. FLAC is compressed, and typically halves the size of the equivalent WAV. It does this the way a ZIP file does: finding patterns and describing them more efficiently, in a way that reverses exactly.
WAV, by contrast, really is uncompressed. It stores every sample as-is, which is why it is so large — around 10 MB per minute for CD-quality stereo, regardless of what the audio contains. Silence takes as much space as an orchestra.
Three minutes of CD-quality stereo:
WAV ≈ 30 MB · FLAC ≈ 15–20 MB · MP3 at 320 kbps ≈ 7 MB · MP3 at 128 kbps ≈ 2.8 MB
FLAC and WAV sound identical, because they contain identical audio. FLAC is simply the sensible way to store it. The only reason to prefer WAV is compatibility — some hardware and software will not read FLAC.
The direction that matters
Conversions between these groups are not symmetrical, and this is the single most useful thing to understand:
Lossless to lossy is a real conversion.FLAC to MP3 orWAV to MP3 takes the full signal and encodes it well. This is what lossy formats were designed for, and at 320 kbps the result is about as good as MP3 gets. You keep the original, so nothing is truly lost.
Lossy to lossless is only a change of wrapper.MP3 to WAV does not restore anything. It produces an uncompressed copy of the already-degraded audio — the same sound in a file perhaps ten times larger. That sounds pointless, and as an upgrade it is. As a compatibility step it is often necessary, which is the real reason to do it.
Lossy to lossy loses a little more each time.M4A to MP3 decodes audio that has already been compressed once and compresses it again with a different method. The second encoder cannot tell which artefacts were in the original and faithfully preserves them, then adds its own. One round is usually inaudible; repeated rounds accumulate.
When you can actually hear the difference
Less often than the debate suggests. Controlled listening tests consistently find that above about 256 kbps, most people cannot reliably identify which file is which, even with good headphones and deliberate comparison.
Where it does show up:
- Cymbals, applause, rain, reverb tails. Noise-like sounds are what lossy encoders handle worst.
- Quiet passages. There is less loud material to mask the encoder's shortcuts.
- After editing. Amplifying, compressing or pitch-shifting lossy audio can expose artefacts that were inaudible before.
- On good equipment, in a quiet room. In a car or a gym, essentially never.
That third point is the practical one. Lossy formats are tuned for a file that will be played as-is. Edit it and you are working with material that has already had detail removed, and the removal becomes audible in ways it never would have been otherwise.
What to actually do
Keep lossless as the master. Recordings, purchased FLAC albums, anything irreplaceable, anything you might edit. A lossless file stays a perfect source forever — you can make an MP3 today and a different format in ten years, and neither affects the original.
Use lossy for delivery. The version you email, upload, or put on a phone. Treat it as disposable: made for one purpose, regenerated from the master when you need something different.
Never delete the lossless after converting. This is the one genuinely irreversible step. Everything else can be redone.
A note on voice recordings
Speech is the exception where lossless rarely earns its space. A voice is a narrow, predictable signal with little for an encoder to damage. A one-hour interview is about 600 MB as WAV and 57 MB as a 128 kbps MP3, and for transcription or reference the difference is inaudible.
Unless you are editing the recording seriously,converting it to MP3 is close to free — and it turns a file you cannot email into one you can.