AAC explained

Advanced Audio Coding is a new audio compression standard intended to replace MP3.

AAC encodes sound in much the same way as MP3. Both methods break down sound into a set of frequencies, and encode how loud each frequency is. This breakdown into a frequency spectrum is called a Fourier transform.

This type of audio compression is very good at representing sound with a "clean" audio spectrum. They will encode flute music and violins very well because these instruments produce only a few distinct frequencies at a time. But sounds with noisier and more complicated spectra, like drum beats, human voice, and mechanical scraping will suffer audibly from this sort of compression.

AAC incorporates a solution to the noise problem in the form of Perceptual Noise Substitution (PNS). PNS comes into play when an AAC encoder encounters noise, such as the sound of wind or an "sh" in speech. Instead of trying to encode the frequency spectrum of the noise, which is difficult, AAC will encode general characteristics of the noise itself.

When you play back an AAC file (anything you buy from iTunes will be encoded with AAC), the playback device will generate sound based on the frequencies encoded in the AAC file. Where noise information is encoded, the device will generate random noise, with the appropriate characteristics, as specified in the file.

The main advantage of PNS is better voice quality at low bit rates. This makes it possible to stream audio with better voice quality over slow internet connections. Also, higher quality voice recordings will take up less disk space.

The trade off is that playing back an AAC file is more work for the playback device. Generating random noise is very computationally intensive. Playback of AAC files on portable devices has only become possible as the computing power of microchips has increased while their size has decreased, as predicted by Moore's Law, to the point that an iPod can be made smaller than a pack of playing cards.

By : Julio Johansen.