http://Science2.FarTooMuch.Info/digsound2.htm is this page's URL.

Digitized Sound - Part 2

Recording to Compact Disc

Red-Book is the digital format is used to store music on Compact Disc. The music is stored as two channels sampled at 44100 times per second with each sample quantized to 16 bits. A value of 32767 represents the maximum air pressure and -32768 represents the minimum air pressure of the recorded sound wave. In other words, the data on the CD represents a plot of the sound pressure with time that contains all frequencies between 20 Hz and 20 kHz with a maximum dynamic range of 96 db. It is common practice to restrict the maximum level of the recorded sound to prevent 'clipping', the artificial limiting of measured levels greater than 32767 to 32767 and less than -32768 to -32768. It is also common practice to add very low level noise to 'dither' the zero reference between plus and minus 1 bit to capture information less the the least significant bit and prevent the generation of digital distortion related to the recorded signal. Correctly done, dither will sound as a very small increase in the uncorrelated noise level. i.e., it will sound as a very low level 'hiss'.

To listen to an audio CD, the data on the CD is first converted to two voltages representing the left and right channels. These voltages are smoothed to eliminate frequencies greater than 22050 Hz. Then these voltages are amplified and used to drive two loudspeakers or each side of a headphone to recreate the recordeded sound that we can then hear. To accurately reproduce all the recorded frequencies requires high fidelity components.

To record sound, the process can be as simple as the reverse process. Two microphones can be used to convert the sound to be recorded into two voltages. These voltages are then amplified to an appropreate level to be sampled. The level of these two voltages is adjusted prevent clipping during the conversion. Clipping is a form of distortion similiar to that created by over driving an amplifier. These voltages are smoothed to eliminate frequencies greater than 22050 Hz to prevent aliasing. In other words, we do not want frequencies greater than 22050 Hz to appear in the recorded data as frequencies less than 22050 Hz that we can hear. To accurately capture all the sound requires high fidelity components.

To get a good recording requires attention to detail in all aspects of the recording process. First of all one needs a place to record that is free of noise. You should not be able to hear the noise of neighbors, noise created by heating or cooling systems, noise created by florescent lights, etc. The recording space should be free of echoes, especially echoes created by parallel walls. Instruments must be in tune. Listen for sounds that should not be their such as a squeekey pedals or keys or chairs.

Chose microphones that complement the music being recorded. Every microphone adds its own coloration to the sound being recorded. Make sure the microphones have the necessary fidelity to capture all the sounds being generatd. Note that the coloration of directional microphones changes with the distance between the microphone and the sound source. The effect of a distance less than a meter can easily be heard. With distances of less than a meter with a blowen instrument or voice, a wind screen should be used. Avoid using microphones with peaks or ripples in their frequency response. This includes both on axis and off axis response.

Often a recording will be made with both loud and soft instruments. This often requires the use of multiple microphones. A recording can be made with different microphone channels mixed together before being digitized. More often, each microphone is recorded to its own track. This has the added benefit of allowing tracks to be added at a later time. In the 'mastering' process, the tracks are mixed down to left and rignt channels when making a red-book CD. It is important that all instruments playing can be heard. Make sure that if the mix contains vocalists or a choir, that the words can be heard and understood. Be very careful when adjusting tone. Excessive tone adjustments decrease a recordings fidelity.

When recording multiple channels, it is good to do the original recording with 24 bits instead of 16. This allows you to maintain a high dynamic range during mixing. If digital effects are to be added, it is good to use a higher sample rate. This helps avoid distortions characteristic of certain 'effects'. After the 2 track 'master' is made, it is then down converted to a 44100 sample rate, if needed, and then to 16 bits. Digital dither must be used to convert from 24 bits to 16 bits.

Then you are ready to put your final creation on CD. After burning your CD, listen to it. Make sure it doesn't have any problems. Listen for noises that shouldn't be there. Listen for digital errors such as skips in time. Listen for distortions such as that created by clipping anywhere in the mastering process. How is the balance between left and right channels? It is good to listen to the CD on "consumer" equipment to check the bass and treble levels. If you are an older engineer, it is reccomended that younger ears listen for high frequency noises that you cannot hear.

You are now ready to make copies of your original work.


This Web Site was designed by Russ Lemon