Audio -- basic technical information

File sizes
This amounts to a tremendous amount of information: 2 tracks * 44,100 samples/second * 16 bits/sample = 1,411,200 bits/second.
CD quality audio stereo, 16-bit, 44.1 KHz = 176 kbyte/sec is too high for CD-ROM (2x drive ~ 200 kbyte/sec sustained ) or modems (28.8
modem ~2.88 kbyte/sec). See Compression below.

A good rule of thumb: Each minute of 16-bit stereo sound at 44.1 kHz requires about 10 Mbytes of disk space.
Thus, with an empty 200 Mbyte hard disk you can record a little less than 20 minutes of CD-quality stereo sound (precisely, 19 minutes, 20 seconds).

Examples:

Possible compromises between sound quality and file size:
Depending on your intended use for the audio, you may be willing to trade some quality in order to reduce the amount of information needed in a digitized sound. Here are some things to consider:

Stereo can often be collapsed into a mono (single track) audio file. If the two tracks are summed, all the sound information will be there, but the directional information is lost. Since computer speakers often are not separated by a suitable distance, even stereo signals are compromised. Going to mono will reduce the file size by half.
Sampling rate: Computers often offer 44K, 22K, 11K, and 6K sampling rates (or numbers very close to these). The sampling rate is a large factor in the sound quality of the digital file. 22K or 44K rates are needed for full-range sounds, while speech is often acceptable at 11K. As you lower the      sampling rate, the sound loses its higher frequencies, so to reproduce the calls of songbirds you may need 44K, but a voice-over may be fine at 11K. To be specific, the sampling rate needs to be twice the highest frequency that is to be digitized.

Bit-Depth: In addition to multiple sample rates, you may be working with either 24-, 16- or 8-bit files.

Lower resolution formats reduce storage space requirements. Unfortunately, lower bit depth and sample rate settings can compromise the audio quality of your sound files. Lower sample rates lose high frequency response, and 8-bit storage causes a reduction of your sound's dynamic range, resulting in noisier, "grainy-sounding" audio, especially during softer passages.

If you are creating 8-bit audio (for example, for multimedia or Internet distribution), you will get best results if you do all your signal processing at 16-bits and 44.1 or 48 kHz, and then create an 8-bit file at the end of the process.  (The quality of 8-bit files can be mediocre.)

You can use an MP3 encoder to convert a file. But its bit-depth is 128 so the file size may increase.

Calculating hard disk free space requirements:
Working with computer-based digital audio requires large amounts of hard disk space. If you are planning on creating new audio files on disk, you'll need enough hard drive space to contain them.

Example: I recorded 13 seconds of sound in the ITC corridor, 16 bit stereo 44,100 hz: This is a 2.5 M file and definitely needs to be compressed.

Audio File Requirements in bytes per second:

number of samples (sampling frequency in Hz)
multiplied by Sample size (1=8-bit,2=16-bit; eg., 8-bit/sample divided by 8 bits per byte)
multiplied by Channels (1=mono, 2=stereo)

A good rule of thumb: Each minute of 16-bit stereo sound at 44.1 kHz requires about 10 Mbytes of disk space.
Thus, with an empty 200 Mbyte hard disk you can record a little less than 20 minutes of CD-quality stereo sound (precisely, 19 minutes, 20 seconds).

CD quality audio (stereo, 16-bit, 44.1 KHz  = 176 kbyte/sec) is too high for CD-ROM (2x drive ~ 200 kbyte/sec sustained ) or modems (28.8
modem ~ 2.88 kbyte/sec).

Arboretum Systems Hyperprism software manual: http://www.arboretum.com/support/manuals/manual_hvst/Files/hppc_digital_audio.html#anchor481814
and a site at Cornell University:  http://www.cit.cornell.edu/atc/materials/dig/avaudio.shtml
and a site at San Francisco State University:   http://msp.sfsu.edu/Instructors/rey/video/bandwidth/filesize.html