Compressing sound: cut what ears can't hear

5 minAS Level1.3.21.3.3

AI-drafted, independently checked · not yet teacher-reviewed

By the end of this lesson

By the end you can describe how a sound file can be compressed, including perceptual coding, and say which methods are lossy and which are lossless.

The MP3 magic trick

CD-quality stereo uses about 1411 kbit/s ( $44\,100 \times 16 \times 2$ bits every second). A typical MP3 uses 128 kbit/s: about 11 times less. Most people can't hear the difference.

The trick: don't store what ears can't hear.

Definition

Perceptual coding

A lossy method of compressing sound that removes the parts of the sound the human ear is unlikely to hear (perceive), so the file is much smaller with little noticeable loss of quality. It is used in formats such as MP3 and AAC, and is sometimes called perceptual music shaping.

What gets removed

Sounds outside the range of human hearing: roughly 20 Hz to 20 kHz (most adults can't reach the top end anyway).
Masked sounds: when a loud sound and a quieter sound of similar frequency play at the same time, the quieter one can't be heard, so it is removed.

The removed data is gone for good: an MP3 can never be turned back into the original recording.

Quick check

Multiple choiceQuestion 1 · Recall

Multiple choiceQuestion 2 · Recall

One-line answerQuestion 3 · Recall

Answer the 3 questions above to continue.