There are two distinct types of sound: 

Synthesised Sound 
Sounds are reproduced using the MIDI (Musical Instrument Digital Interface) via a sound card in the computer. This card contains a chip which accepts a set of music producing commans. These commands specify the instrument, duration and note so that it can reproduce the sound. This approach is very suitable for reproducing sounds that can be reproduced by sound cards, rather than voices and sound effects. 

Sound Data 
A digital recording of an analog signal. What we do is, we are sampling the streaming sound with a specified interval. If we want the quality to be high, then we should sample frequently. As a result, the higher the quality of the recording, the larger the file. Common rates are 11, 22 and 44 KHz. Each sample uses 8, 16 or 32 bits.

Since the file contains a list of commands for your soundcard to interpret, large pieces of music can be stored in relatively small files. 

Hence, the reproduction quality is very good

It can only reproduce musical sounds, so is unable to be used for sounds such as voices, or sound effects like a dog barking or glass breaking.

a set of sample data and  
a description of how the samples are to be played

WAV: Also called RIFF WAVE, Wave Form Audio is the MS Windows standard. WAV files are sound data files, and so contain a digital representation of analog signal. The greater the recording quality, the larger the file, and WAV files have a habit of becoming huge. For example, an 8-bit recording can use 1.5 megabytes per minute, and 16-bit can consume 3 megabytes per minute. 

AU: The original NeXt computer standard, and relatively popular on the internet. It is a similar format to WAV, and so also uses a lot of disk space. 

MP2/MPA: These are MPEG audio files, which are sound data files like WAV, but use a compression routine to shrink the file size. Like JPEG image files, MPEG audio leaves out information that is usually outside the auditory perception of the listener. Because of their size and excellent quality, this format has gained popularity on the internet, although they use a lossy compression scheme. 

RealAudio: This is an emerging format specifically designed for the Internet. RealAudio compresses a standard audio file (such as a WAV) and sends it as a stream of bits to the user. This means that, instead of waiting to download the entire file before you can hear it, as for the formats above, RealAudio plays back as it is received. However, what is gained in immediacy is lost in sound quality, although recent developments in this format provide a quality equal to the radio. Quality suffers most when there are many similtaneous sounds. But because of 10 megabyte WAV file can be compressed to less than 300 kilobytes, RealAudio gives the ability 
to transmit program length sound across the Net.