What Is Audio Compression?(There Are Two Answers)

Why this term is genuinely confusing

"Compression" in audio can be discussing a file format, or it can be discussing a signal processing technique. These two uses of the word exist in completely separate contexts — one is about file storage, the other is about sound shaping — but they collide constantly in casual conversation.

A recording engineer saying "I compressed the vocals" means they used a dynamic range compressor to smooth out the performance levels. A developer saying "the audio is compressed" typically means it was encoded as MP3 or AAC. Both are using the word correctly. Both can confuse someone from the other world.

Understanding the difference matters because the implications are completely different. Data compression affects file size and potentially audio quality. Dynamic range compression affects how the audio sounds — the feel, the loudness, the punch.

Side by side

Data compression: how it works

Data compression takes audio data and finds ways to represent it using fewer bits. The result is a smaller file. How much smaller depends on the codec and whether it's lossy or lossless.

Lossy data compression uses psychoacoustic models — it removes sounds you're unlikely to notice, like frequencies masked by louder sounds, or detail above the upper range of human hearing. At high bitrates, the removal is subtle. At low bitrates, it becomes audible. Lossless compression finds redundant patterns in the audio data and stores them efficiently — like ZIP compression, but for audio.

Dynamic range compression: how it works

Dynamic range refers to the difference between the quietest and loudest parts of an audio signal. A live orchestra might have a dynamic range of 70 dB or more — whisper-quiet passages and ear-splitting climaxes. A radio broadcast is compressed to about 10 dB — everything sits at a similar level.

A compressor is a device or plugin that automatically reduces the volume of audio that exceeds a set threshold. It brings peaks down toward the average, which makes the overall audio feel more controlled and consistent. Four key controls:

• Threshold:The level above which compression kicks in. Sounds louder than this get turned down.
• Ratio:How much the compression reduces levels above the threshold. A ratio of 4:1 means every 4 dB over the threshold becomes 1 dB.
• Attack:How quickly the compressor reacts to loud sounds. Fast attack catches transients; slow attack lets them through.
• Release:How quickly the compressor stops reducing volume after the loud sound ends.

Dynamic range compression is used everywhere: on podcast vocals to smooth out inconsistent speaking levels, on music to punch up drums, in broadcast to meet loudness standards, and in mastering to prepare final audio for streaming platforms.

How to tell which one is being discussed

Context is usually the giveaway. If someone is talking about file formats, downloads, encoding, or file size — they mean data compression. If they're talking about mixing, mastering, plugins, signal chains, or how audio sounds — they mean dynamic range compression.

The overlap: when both apply

A finished, distributed audio file has typically had both types of compression applied — just at different stages.

Dynamic range compression (and limiting, its harder-edged sibling) is applied during mixing and mastering, shaping the audio signal. Then the finished master is encoded as an MP3, AAC, or OGG file — applying data compression.

Streaming platforms add another layer: they measure the loudness of uploaded audio and apply normalisation to bring everything to a consistent level — effectively a mild form of dynamic processing — and then serve it in a lossy format like AAC or OGG.

Understanding both types helps you understand why audio sounds the way it does at every stage of the chain.