Digital Recording

This article is intended to be a high-level overview of concerns related to digital audio recording on a personal computer. Subtopics in this article may be split into separate articles in the future, but for now consider this a one-stop shop.

DAW

A DAW (Digital Audio Workstation) is a computer system capable of recording and processing digital audio. When someone refers to recording into a DAW, it means they are storing the audio on a hard disk rather than using Analog Tape which before the advent of Computers was the primary way to record.

"DAW" is a fairly wide-ranging term, and can mean anything from a dedicated system such as the Roland VS series workstations to a standard personal computer with an audio interface and software capable of manipulating audio.

Audio Interfaces

Getting the best quality of audio recording requires a quality audio interface, with a stable clock and good converters. Many audio interfaces also include one or more microphone preamps. The audio interface also determines how many channels of audio can be simultaneously recorded. Most common interfaces use the PCI bus, Firewire, or USB.

Audio interfaces range in price from approximately $200 up to several thousand dollars. More money typically buys you more I/O channels, more I/O formats (analog, SPDIF, ADAT, etc), more or better preamps, better converters, and/or a better clock.

If you are planning on recording drums or other instruments that require a large number of microphones, make sure your audio interface has enough channels to support this. Also be aware of what type of inputs there are on the interface. If you want to record drums, you might need 8 inputs. The unit might advertise 8 inputs, but that could be 2 S/PDIF, 2 analog line ins and 4 channels of ADAT. Therefore, you will need an outboard converter for the ADATs, pre-amps for them, pre-amps for the analogue ins, another converter for the S/PDIF, and mic pre-amps for that too. That's liable to cost you a lot. If you don't already have a mixing desk, you will want plenty of pre-amps built in, or reserve some cash to buy some. If you have a desk but no converters, a load of line ins will do. If you have converters, from, for example - an old ADAT-based setup, a setup comprising only of ADAT inputs would be fine.

A quick run down of the types of inputs are below:
Mic Pre-Amp: The input consists of an XLR input, and a mic pre-amp like you would find on a mixing desk. There will usually be a gain knob on the front of the unit. Users without mixing desks or standalone pre-amps will find these ideal.
Line In: The input consists (usually) of a jack input, (but could be xlr, or d-sub), and will need an external pre-amp to connect a microphone. This could be in a desk, a standalone pre-amp; or you could use the line-in for a line in such as a sampler, synth or the return from an FX unit. Users with existing analogue studio setups converting to digital will find such units ideal.
ADAT (Lightpipe): The input consists of a small connection which can take a "lightpipe" cable, which can usually carry 8 channels. These will come from an outboard converter, such as an ADAT unit, or the output of a digital desk. You need therefore an outboard converter, and also pre-amps to feed the converter, or a unit containing both such as the Focusrite Octopre. This connection would be ideal for anybody working from an existing ADAT-based studio.
S/PDIF: Sony Philips Digital Interface. This takes a phono-looking cable (looks like something used in hifi) which can carry 2 channels of digital audio, and also a word clock signal (unlike lightpipe). Works similar to the above, with external converters and pre-amps being required.

If you find that the unit you have does not have enough inputs for what you want, it doesnt mean the end of the line for the interface. Most audio applications and operating systems allow you to run 2 devices simultaneously, so you can always add another interface to your existing setups. For companies such as MOTU, it is very common to see 2 or more of their firewire units connected together via firewire to extend the number of available inputs.

Converters

Getting audio into and out of a DAW requires hardware to convert the audio from an analog waveform to a digital signal, and vice-versa. Converting analog to digital is called AD conversion, while digital to analog is called DA conversion. Any dedicated audio interface will of course include converters.

Converters are a critically important part of digital recording, as they can have a great impact on your overall sound. It's important to use an interface with high quality converters, as the best microphones and preamps available won't sound great if subjected to poor conversion.

Digital Clocks

A digital clock is an oscillating chip built into most digital audio devices which synchronizes the device and, optionally, synchronizes other devices. A stable, reliable clock is a requirement for great digital audio. A poor clock source can cause clicks, pops, loss of stereo width, frequency issues, or other audio quality problems.

A stable clock becomes even more important when running multiple digital devices together. Though devices may have their own internal clock for running standalone, many digital devices support synchronization with an external clock source so that the entire audio chain is synchronized.

Indeed, there are several devices on the market that are nothing but clock sources, such as the Apogee Big Ben and the Drawmer D-Clock, which are meant to serve as the master clock for an entire digital studio.

Operating Systems

Both Macs and Microsoft Windows have a wide range of DAW applications available and are commonly used. Linux and other BSD based operating systems are not widely used in audio recording both because of the general public's unfamiliarity and lack of major vendor support; however, Ardour (OS X, Linux) is a fully capable audio editor for linux and may increase its use in studios.

Linux Distributions Targeted at Audio

Here are a few (but probably not all) distributions targeted specifically for audio (and sometimes video and graphics) users.

Linux Sound list of audio distributions

Some of the more common DAW application are:

Windows Only
Adobe Audition (formerly Cool Edit Pro)
Cakewalk Sonar
Magix Samplitude
Magix Sequoia
SAW Studio
FL Studio Pro
Acid Pro

Mac/Windows
Ableton Live
Digidesign Pro Tools HD
Digidesign Pro Tools LE
Mackie Tracktion
Steinberg Cubase
Steinberg Nuendo
Propellerhead Reason
Ardour (OS X/Linux)
Audacity (OS X/Linux/Windows)

Mac Only
Apple Logic Pro
MOTU Digital Performer MOTU's Digital Performer Website

Sample Rates and Bit Depth

Digital audio is a computerized representation of real-life sounds. In order to be processed on a computer, analog audio must be converted into digital data. The quality of the audio is determined by the sample rate and bit depth at which the audio was captured. The higher the sample rate and bit depth (within reason!), the better the quality of the audio.

Sample Rate refers to how many times per second the audio information is captured. Sample rates are measured in Kilohertz (KHz). 1 KHz equals 1000 samples per second. So, audio recorded at 44.1 KHz captures data at a rate of 44,100 samples per second. Most audio interfaces support sample rates of 44.1, 48, 88.2, and 96 KHz. Many higher-end devices also support 176.4 and 192 KHz sample rates as well.

Bit Depth refers to how much information is captured with a sample. Each sample measures the amplitude (shape) of the wave, and higher bit depths allow the analog shape to be more closely matched. Imagine a piece of graph paper, on which you were trying to line up the graph with the smooth analog curve of a sine wave. The smaller the boxes on the graph (higher bit depth), the more closely you could match the shape of the sine wave.

In the diagrams below, the X axis represents sample rate (each line equals one sample) and the Y axis represents bit depth.

At a low bit depth, the shape of the wave cannot be matched very closely	At a low sample rate, the shape of the wave also cannot be matched very closely
At a higher sample rate and bit depth, the shape of the wave can be more closely matched.	However, the higher the frequency of the sound, the more difficult it is to closely match the wave.

Bit depth is exponential, so a 24-bit sample holds 256 times as much information as a 16-bit sample. A 16-bit sample has a range of 65,536 values, while a 24-bit sample has a range of 16,777,216.

Nyquist Theorem

The Nyquist Theorem (named for Harry Nyquist) states that the highest frequency that can be accurately reproduced is not more than half of the sampling rate. For example, a CD (at 44,100 Hz) could theoretically reproduce frequencies of up to 22,050 Hz.

One qualm with the theory argues that even though the human ear cannot "hear" anything above approximately 20 Khz, the sampling inaccuracy of the higer frequencies will color the sound in subtle ways.

Also, the accuracy of the 20hz signal in 44.1 is fairly "low quality" when taking into account the quantizing that occurs.

In Practice

It is generally agreed upon that 24-bit is a sufficient bit depth for just about all audio recording. However, there is much disagreement about what is the "best" sample rate to use. Some argue that anything above 96 KHz is overkill, while others claim to be able to hear a difference in audio quality at the highest (176.4/192 KHz) sample rates.

Whatever you decide to use, be aware that increasing bit depth and sample rate have two negative effects. First, it increases the size of your audio files. Second, it increases the load on your computer. A computer that is perfectly capable of handling a large project recorded at 44.1 KHz may not be able to handle a similar project recorded at 96 KHz.

External Links

Wikipedia's article on the Nyquist Frequency

Categories: Technical | Software