Front Door

Pricing

online sales

Links

Contact Us

Products

Tech Support

How to Order

find out about Authentic Sound Recorder or Authentic Sound Studio.

Digital Recording Concepts

Our plan is to use this page to explain and illustrate basic concepts from the field of digital recording.

Send comments regarding this page to the webmaster.

 

 

 

Sample Size: How Many Bits are Enough?

 

Most digital measurements, and on a purely technical level recording sound must be thought of as a measurement as well, involve the translation of some form of energy (such as sound) into an electrical signal. This electrical signal is usually expressed as a Voltage or electrical potential difference. The incoming voltage is an analog signal in that it has not yet been mapped into a binary number or series of binary numbers. It is through this mapping process that the number of bits used becomes important.

A common word size is 8 bits and the largest number that can be represented in this form is 255 (=27 as the lowest order bit represents the coefficient of 20=1). The incoming analog signal is a Voltage somewhere within some range of Voltages that the sound card or other analog to digital converter can accept. A common range is 0 to 5 V DC. With an 8 bit word the voltage range is divided up into 255 intervals each representing about 0.02 Volts.That means that Voltages separated by less that 0.02 Volts will often be read as equivalent.

You might think that increasing the sample size to 16 bits reduces this effect by a factor of 2. The difference is much more significant as 8 more powers of 2 are now available. With 16 bits the largest number that can be represented is 65536 (=215,quite a bit larger than 255) and thus a 5 Volt range can now be divided up into segments roughly 0.00008 Volts apart. The measurements now are much more sensitive to subtle variations in the Voltage stream resulting generally in a less muddled signal. There is also less chance of overflows and therefore less harmonic distortion.

So for twice as much storage you can,at least in principle, significantly increase the quality of the digitized signal.

top

Fourier Spectrum: Why is it important?

You may never need to look at a Fourier spectrum but its importance in recording and especially editing is hard to challenge. What is a Fourier spectrum anyway? The basic concept is not too hard to grasp. The incoming Voltage or bit stream (tot he sound card) is said to exist in the time domain in that distinctions are made between readings based on when they occur in time. Fourier (a French mathematician) discovered a way to represent any signal in the time domain as a signal in the frequency domain. That is, instead of distinguishing the data in terms of when samples occur in time you look at the fraction of the total signal at each frequency over some range.. If you include a large enough frequency range the time domain signal is completely represented in the frequency domain, so no information has been lost, it is just being represented differently. The process of going back and forth between the time domain and the frequency domain is the subject of Fourier analysis. using computers most Fourier analysis is performed using a numerical algorithm referred to as the Fast Fourier Transform or FFT.

Why is this so important in sound recording? Because sound conveys information (speech, music etc.) and important information is encoded within the frequency spectrum of a sound wave. For example, the pitch of a note varies with the frequency of the incoming sound wave. Using the Fourier spectrum in the editing process allows for a direct editing of the frequencies present in a signal. So processes like filtering are much easier.

top

Reverb and Convolution

In a sense, reverb is a form of echo. Both reverb and echo occur when signal from the past mixes with signal in the future. Echos generally come back only once, where as reverberation (reverb) results when signal from the past reflects multiple times (off walls etc.) and mixes repeatedly (with diminishing strength) with signal in the future. Each produced sound lingers.

The key to artificially creating reverb lies in figuring out a way to allow each individual sound sample to not fade away (so to speak). more about this topic

top

Latency while Recording

Latency usually refers to the delay between the time an analog signal enters a sound card, and the time it is heard through the speakers or headphones attached to the computer. But it can also refer to any delay experienced while engaged in digital recording and/or playback of tracks. You will have a lot of problems with latency if you use your computer as a monitor. For example, if you are trying to record the signal from an electric instrument that is sent directly into the sound card. You will almost always experience a delay between the actual production of a note (plucking a string, pressing a key etc.) and hearing the sound. This will cause problems with synchronization if you are recording while other tracks are playing and probably make it hard to play. The delay occurs while monitoring because of the time required for the sound card to digitize the incoming signal record it and then convert the now digitized signal back to analog so it can be sent to the speakers or headphones for playback.

It is commonly believed that anything digital is faster than its analog counterpart. The latency problem illustrates how this is not always true. Electrical signals travel through circuits (digital or analog) essentially at the speed of light and will transverse most analog circuits with no noticible delay. Digital circuits being significantly more complex than analog circuits, hang on to a signal longer and in certain types of applications (e.g. digital recording) a delay can be observed. Even with the latest high speed processor (say a G4) there is still some latency, and unfortunately the problem cannot be completely eliminated by beefing up your system with more RAM or other improvements. Hopefully the problem will be eliminated in the future, but for now we'll have to live with it.

Solution for Now:

(1) If you are having trouble with latency, try to find a way to either externally monitor your signal or use a microphone instead of direct input.

(2) If you are using AS Studio you have two options for dealing with latency:

(a) When you mix use the Mix Sequencing window to advance the starting time for the base track in your mix as it is probably ahead of the others. You probably need to advance it by about 20-40 ms.

(b) Use the Carbon version under OS X. There appears to be much less latency eveident under OS X.

Here's a good article from Sound on Sound magazine about the latency problem.

http://www.sospubs.co.uk/sos/apr99/articles/letency.htm

 

top