With the emergence of relatively low-cost, high-quality portable digital recorder's, it has never been easier for amateurs to record high quality digital audio from almost any source, in almost any environment. All that one needs to obtain semi-professional results is a basic understanding and application of the following major points:

1. Bit-depth and sampling rates
2. What is necessary to record high-quality digital audio
   1. Microphone type, model, and placement
   2. Cables and connections
   3. Microphone pre-amp quality
   4. Mixer quality
   5. Digital recorder pre-amp quality
   6. Gain levels at each input
   7. Master recording medium/bit-depth & sample rate
3. Audio file distribution and compression

Because digital audio is sampled (i.e., the recorder/playback device takes samples of sound at regular intervals, to see "what it's like" at distinct points in time), the first thing you need to understand is what's involved in digital sampling.

If you stand outside a voting booth and interview every 10th person who votes, you'll end up with a set of samples that gives you some idea of what the outcome of that vote will be. If you interview every 5th person who votes, you'll end up with an even better idea of the outcome. If you interview every 2nd person who votes, you'll end up with a very good idea of the outcome. And if you interview everyone who votes, then, well... you'll end up with an exact idea of the outcome! That's why we don't use exit polls to determine who wins an election--we count all of the votes, instead. Except in Florida.

In this example, the more people you sample (i.e., the more often you sample voters), the more accurate your exit poll results will be. But also, the quality of your poll results depends on how much information you gather from each voter. For example, you'll gather more information, and more relevant information, when you ask voters which candidate they voted for, rather than which party they voted for.

The same principle is true with digital audio sampling. The more often you sample the audio as you record it/play it back, the more accurate your recording will be at representing every nuance of the original sound. But at the same time, the amount of specific information that you gather about the sound each time you sample it also has a big impact on the quality of your results.

Digital audio sampling has two important characteristics: bit-depth and sampling rate. The bit-depth determines how much information about the sound is gathered every time you take a sample of the sound. And the sample rate determines how often you sample the sound.

Obviously, bigger is better in the case of digital sampling, because the more information you gather about each sample, and the more often you sample the sound, the more accurate your final results will be.

However, you cannot gather an infinite amount of information about the sound each time you take a sample. Likewise, you cannot gather an infinite number of samples, either. The buck has to stop somewhere. There are practical limitations set by the technology itself, in terms of computer processing power that is required to create these samples while recording, and to decode those digital samples back into audio when you play them back.

So your goal when recording quality digital audio is to find the best compromise between the bit-depth (the amount of audio information that is recorded each time you sample), and the sample rate (how often you take a sample of the sound), to reproduce the sound as accurately as possible given the limitations of the recording and playback technology you have available.

The Compact Disk "standard" is to sample and play back sound at depth of 16 bits, and a sample rate of 44.1KHz (KHz = kilohertz, or thousands of cycles per second). This means that the amount of information you can capture about a sound during each sample is represented by 16 bits, and how often you take a sample of the sound is 44,100 times each second.

Let's look at each of these components individually.

Computers (i.e., digital devices) store and process information using the binary number system, which has only two digits, 0 and 1. When you string these digits together, each digit is called a bit, and each bit can only be either a 0 or a 1, since the digits "2", "3", "4", "5", "6", "7", "8", and "9" don't even exist in the binary system.

In our decimal (base 10) number system, one "bit" can represent any of ten possible states: 0, 1, 2, 3, 4, 5, 6, 7, 8 or 9. Two "bits" in our decimal system can represent any of one hundred possible states (00..99). In the same way, three "bits" in our decimal system can respresent any of one thousand possible states (000..999). And so on.

Notice that the number of possible states that can be represented by X number of bits in the decimal system is always 10 to the power of X (10^x).

Well, why should it be any different in the binary number system? The only difference is that binary is a base 2 system, rather than a base 10 system. So X number of bits in the binary system is capable of representing one of 2^X possible states.

Since the Compact Disk standard is 16 bits, this means that each sample represents one of 2¹⁶ possible states. That is 65,536 possible states. Each time audio is sampled at CD quality, the computer has to decide which of those 65,536 possible states most closely represents the sound, relative to all the other samples that are taken. This may sound like a lot of choices, but think about how complex and magnificent every sound we hear really is! Think about how different a piano, a violin, a crying baby, an opera singer, a motorcycle, a jackhammer, the wind, a scream, and a whisper all sound. At some point, 65,536 choices to distinguish all those sounds from one another at a distinct point in time begins to seem rather limiting, doesn't it?

The new generation of digital recorders are able to record audio at 24-bits. This means that each sample can represent one of (2²⁴) possible states. That is 16,777,216 possible states, at 24-bits.

So 24 bits can represent more than sixteen million possible states, versus the mere sixty-five thousand possible states that can be represented by 16 bits. Now it finally seems possible to accurately represent the difference between a crying baby and a jackhammer, right?

Well, that's true. However, due to our brain's ability to "fill in the gaps", the CD-quality standard of 16 bits is actually good enough to accurately represent the difference between a crying baby and a jackhammer--on one condition: you have to sample the sound very often in order for this aural illusion to work. If you don't sample the sound often enough, the resulting audio will sound mechanical, or like it's breaking up.

In the case of the CD standard, you must sample the sound 44,100 times each second. And that's pretty fast! This is why quality digital audio was not possible before the development of computers that operate at very high speeds.

The typical human can hear high-pitched frequencies up to around 20KHz (20,000 cycles per second). For reasons that are beyond the scope of this article, the sampling rate of digital audio is effectively twice the highest pitch frequency that it is capable of sampling. So 44.1KHz (44,100 times per second) can sample frequencies up to 22,050KHz, which is an even higher pitch than humans can hear. In other words, it really doesn't serve any practical purpose to sample sound faster than the 44.1KHz CD-quality standard, because humans can't hear those higher frequencies, anyway.

So when you combine the ability to represent a sample with one of 65,536 possible states with the ability to do so 44,100 each and every second, the result is the audio equivalent of looking at a newspaper photo that is made up of very tiny pixels. At a casual glance, you can't tell the difference between a newspaper photo and the original black and white film photo. And at a casual listening, you can't tell the difference between 16-bit, 44.1Khz digital audio and the original analog sound. That's why some people, somewhere, at some time, got together and decided to set the compact disk standard at 16-bits/44.1KHz.

Since 16-bits, 44.1KHz is all we really need to represent sound accurately in the digital world, you may wonder why companies produce digital recorders that can record at 24-bits, with sample at rates of 48KHz, or even 96KHz? Well, the short answer is that they do so because they can. Those specifications sound pretty impressive, and they help to sell products. And the truth is, the sound quality is better. The question is, can the typical human being really tell that it sounds better? And if so, how much difference do they really notice?

In the real world, most people do not notice any difference at all. Some people may claim to notice a big difference. But then again, some people claim to have been abducted by aliens. In terms of the science involved, however, there really isn't any practical need to record at better than CD quality, for everyday purposes. Instead, if you're recording digital audio at the CD standard of 16-bits, 44.1KHz, the main factors that will determine the quality of your audio are the the equipment you use to make the recording, the equipment you use to play that recording back, and the skill you exhibit as a sound recordist. So those issues are what the remainder of this essay are concerned with.

In terms of the equipment and techniques you need to record high-quality digital audio, you merely need to consider the quality that exists at each point from the sound source to the recording medium, beginning with the microphone(s).

A microphone is a device that converts audio sound into an electrical signal that accurately represents the original audio sound. There are several effective ways to accomplish this, so there are also several basic types of microphones. The most common types are dynamic, and condenser microphones.

Dynamic microphones produce a relatively strong electrical signal that is powerful enough to be recorded directly, without the need for amplification first. Condenser microphones produce a much weaker signal, and they usually require some kind of pre-amplification before the signal will be powerful enough to record, or to manipulate with special effects.

Because of the way they're constructed, dynamic and condenser microphones have very different audio characteristics, so you must decide which type of microphone is most appropriate for the sound you are recording. For example, dynamic mics are usually more focused, and less sensitive than condenser mics. This means that dynamic mics must be positioned carefully to pick up the specific sound source that you wish to record, but once you've positioned them properly, they won't pick up an unreasonable amount of sound from other nearby sources. Condenser microphones are very sensitive, so even when they're positioned properly they will still pick up a lot of extra sound from other nearby (or even distant) sources.

Sometimes you want the extra sensitivity that a condenser mic provides, like when you seek to record ambient noise to place the main sound source in a proper context (i.e., children playing in the background while interviewing a day care provider). However, the main appeal of condenser microphones is their frequency response and audio fidelity. A good condenser microphone captures more fullness and body than a dynamic mic does. So dynamic mics are often used for purposes where the sound source is focused, and tends to fall within a relatively narrow range of frequencies (like a singer in a rock band), while condenser mics are used when you want to faithfully reproduce the full spectrum of the source's sound (e.g., an interview, or film soundtrack).

If you hand a rap musician a condenser microphone, you won't like the results. Rappers hold microphones close to their mouths to maximize the the force of their voice, while minimizing background noise from the music itself. A condenser microphone would pick up all the background music, and would probably be blown out by the rapper's forceful voice. So rappers (and stage vocalists in general) tend to use dynamic microphones, instead.

On the flip side, if you put a dynamic microphone on David Letterman's desk, it will fail to capture the full body and depth of his voice. And because dynamic mics are more directional, the audio will sound weak and flat whenever Letterman moves to the left or right, or moves too close or too far away from the microphone, rather than speaking directly into the microphone from the same position. Condenser mics are the preferred microphone type for this kind of situation, where there is relatively little background noise, and the audio source tends to move around.*

* Technically, Jay Leno and David Letterman's voices are captured by highly-focused shotgun microphones that are mounted above their heads, off-camera (or by stage hands holding a boom with a shotgun mic mounted on the end). The large, radio-style microphones on their desks are merely for decoration. But the principle remains the same--these shotgun mics are still condenser mics.

Another important distinction between microphone types is unidirectional ("directional") versus omnidirectional ("omni"). Directional mics pick up sound from a relatively narrow field, usually right in front of the microphone. Omni mics pick up sound from a much wider field, and they may even pick up sound 360 degrees around the microphone. Depending on the background noise that exists, and how much of that background noise you wish to record, using the wrong microphone type here can dramatically degrade the quality of your recording.

After you've decided on the type of microphone that best suits your situation, you need to choose a brand and model. Different models from different manufacturers have different audio characteristics, like frequency response, signal to noise ratio, and polar pattern (a graphic representation of the locations where the microphone picks up sound). The general rule "you get what you pay for" applies here. However, you can get by in most situations just fine using microphones that cost anywhere from $75-$300. A microphone that costs $10-$50 will probably introduce a lot of noise, and fail to capture the full range of frequencies that you wish to record. Yet, there also comes a point where you're paying a lot of extra money for nothing but pretense. So there is no reason to believe that you need. a $1000 or $10,000 microphone to produce superior recordings. If all the equipment you use is low-noise and high-quality, you can easily get by using even a $100 microphone.

After you've chosen your microphone(s), it's important that you position them properly relative to the sound source. For example, if you place the microphone too close to a loud source, your recording may be heavily distorted. But if you place it too far away, the recording may lack the power and presence of the original sound source.

The unique acoustics of your recording environment will also affect your recordings, and you need to account for this when placing your microphones. For example, if you're recording an orchestra in a large hall, placing the microphones in the balcony will capture not only the performance, but also the reverberation of that performance throughout the entire hall (not to mention the audience's coughing, sneezing, talking, etc.). In this situation, it is better to place the microphones near the orchestra itself, so your recording is dominated by the main performance only. However, at a rock concert, you may want to move the microphones back quite a distance, to capture some of the amorphous audience noise that helps to give the performance a truly "live" feel.

Microphone placement is especially important when you record in stereo. In fact, it's crucial when you record in stereo. Several well-established methods of stereo microphone placement have already been developed, and each method has its own pros and cons. So do some research on microphone placement, and run some tests with the microphones you'll be using before you begin making that important recording you've been preparing for all along. That way you'll have a good idea of what to expect from your results before you begin, and it's less likely that you'll experience unpleasant surprises later on in the listening room, when it's too late.

An audio cable is a wire (or a combined/twisted set of wires) that transmits an electrical (audio) signal from one device to another. All things being equal, there's no practical difference between a $5 cable and a $75 cable, except for the price, because, again, the purpose of a cable is to transmit an electrical signal from one device to another, and this is something that cables either do (if they work), or don't do (if they don't work).

Some cable manufacturers have spent a lot of time and money on marketing (rather than product development) to establish their brands of wire as somehow being higher quality than other brands of wire, by selling space-age looking audio cables in pretentious packaging at unconscionably high prices. Nonetheless, whether it costs $5 or $500, if it transmits an electrical signal from one device to another, then the cable has done it's job, and has done it's job well, and you should be proud of it. So there's no reason to spend $30 or $50 or $100 on a cable (i.e., a piece of wire), when a $5 cable will do just fine.

However, with that said, it is important to ensure that your cables are well shielded (insulated), so they don't pick up electronic noise from nearby devices like computers, monitors, and cell phones. It's also crucial that your cables do not suffer from internal shorts, and that there's a good, clean connection to the jacks at each end of the cable.

But then again, these are merely the characteristics of cable that does what a cable is supposed to do. There's nothing magical or mysterious about it. So again, there's no good reason to pay $30 or $50 or $100 or $500 for a cable, just so you'll look cool. All that extra money is better spent on quality microphones, pre-amps, and recording devices, instead.

Because again, an audio cable is just a piece of wire. That's all a cable is. Have I made my point yet?

There are also various adapters that you might need, in order to connect devices to each other. Common adapters convert 1/4 inch TRS plugs into 1/8 inch (3.5mm) plugs, or vice versa, and these adapters come in both mono and stereo versions. Other important adapters convert 1/4 inch TRS to a 3-pin XLR, and vice versa.

The most complicated part about dealing with cables and adapters are the concepts of "balanced" (3-pin XLR) versus "unbalanced" (TRS, 1/4in.), and impedance. Differences in impedance may require the use of an impedance-matching transformer. In any event, the purpose of these terms is to describe electrical phenomena that are analogous to heavy city traffic. When traffic is tight and congested (e.g., a traffic jam), the cretin who blows down the highway, weaving in and out of traffic at high speed, is nothing but a dangerous menace who threatens to kill someone someday. A traffic jam is like a high-impedance circuit, and plugging a low-impedance cable or device into that circuit can be a dangerous menace, and it may even kill one of your devices. At minimum, failing to take these issues into consideration can introduce a lot of noise, hum and hiss into your recordings. These concepts are not as complicated as they appear to be, but it's crucial that you do at least a little bit of research on them, to protect your equipment, and to maximize the quality of your audio recordings.

A similar issue is the difference between a microphone jack, and a line-input jack. A microphone jack is meant to accept the relatively low signal that a microphone produces, and that signal is usually amplified by a pre-amplifier circuit that is connected to the jack. A line-input does not have a pre-amplifier. Instead, it assumes that you will set the output levels on the device you plug into it, in order to get the signal power you're seeking. Since microphone inputs have pre-amps, and line inputs do not, line-inputs are usually much quieter than microphone inputs (i.e, they introduce less noise into your signal).

Even though a condenser microphone's inherently low signal requires some kind of pre-amplification, there is nothing that says you can't also use a pre-amplifier with a dynamic microphone, in order to "color" the sound from the dynamic mic. However, keep in mind that condenser mics typically require 48volt phantom power to accomplish anything at all, yet that same 48volt phantom power might destroy a dynamic microphone.

In any event, the quality of the microphone pre-amp that you use--if you use one--seriously impacts the quality of the final recording. If the mic pre-amp is noisy, that noise is only going to be magnified as the signal travels through other equipment on its way to the final recording. However, if the pre-amp's output is clean and noise-free, you will only have to worry about noise generated by other devices downline.

There are two basic types of pre-amp circuits: tube amps, and solid state (transistorized) amps. Tube amps use a vacuum tube to accomplish the signal amplification. Tube amps are analog devices, while transistorized amps are digital devices. The advantage of a tube is that it produces a warm and natural-sounding signal. The disadvantage is that tubes inherently generate a fair amount of noise and hiss (and heat!). Transistorized amps can sometimes be very quiet, indeed, yet they also tend to produce a more dry and sterile sound that lacks warmth, and may sound less-than-natural.

Some microphone pre-amplifiers, like the Presonus TubePre, feature both a tube and transistorized circuits. This combination allows you to mix the two types of circuits together to maximize the good, and minimize the bad, for each unique recording situation.

If you use a mixer, be aware that despite all those physical buttons, knobs and sliders, it is still an electronic device, and therefore it can potentially add noise to your signal. So the quality of a mixer's pre-amp and internal circuitry is important. If the mixer's pre-amp and other electronics generate a lot of noise, then the recording device's pre-amps will only amplify that noise later on. In general, it's best to avoid using a mixer altogether for general-purpose recording, since no "mixing" of multiple signals (the main purpose of a mixer) is really necessary in these cases. However, if you're miking individual instruments in a band, you will need a mixer to combine all those signals into a stereo or mono signal to send to your recording device.

Recording devices usually have their own pre-amps to boost signals that are input to the microphone jacks. However, a digital recorder has a very specific function: to record an audio signal onto some kind of medium for future playback. Amplification is not a part of this essential function. Because of this, low-cost digital recorders are notorious for including low-quality pre-amps that can take a nice, clean signal from your microphone/pre-amp/mixer chain, and turn it into a signal that hisses and spits like an angry cat--at the very last stage before you record that signal!

So if you are using an inexpensive recorder with poor-quality pre-amps, you want to send the strongest signal possible to your recorder, so you don't have to turn the gain up high on the recorder's internal pre-amps.

The good news is that more expensive digital recorders today usually feature anywhere from decent to excellent-quality pre-amps at the microphone inputs. This is one reason why you get a good return on your investment when you spend the extra money to purchase a recorder that has quality pre-amps. After all, the actual act of recording the signal onto a medium isn't all that complicated--even a cheap digital voice recorder or mp3 player with voice-recording capability can accomplish that.

The line input on your recorder may have no pre-amp at all. In this case, the input gain is fixed, and you rely entirely on the previous device(s) in the chain (e.g., microphone/pre-amp/mixer) to send an appropriately strong signal to the recorder's line input. This is definitely the input to use when you can provide the recorder with a strong, clean signal from the microphone, because it eliminates one potential source of noise (the recorder's pre-amps) from the chain altogether.

In audio electronics, "gain" refers to the power of the audio/electrical signal. Many people confuse "gain" with "volume". But volume is determined by the power of a signal sent to some kind of reproduction device, like an earbud, headphone, speaker, or PA horn. Gain refers to the power of the signal itself as it exists throughout the audio chain, even where no reproduction device exists.

It is very important to set the gain carefully at each input of each device in the chain. The original gain is determined by your microphone's output. That's why some microphones are called "high-gain" and others are called "low-gain" microphones. Either way, the microphone's design and construction alone usually determine the microphone's inherent gain*. After that, every time you plug another cable into another device, you are the one who determines the input gain, by manipulating the input (and possibly output) gain controls on each device.

* Some microphones do include a gain control that allows you to manipulate the microphone's output signal level.

Setting the gain levels properly on each device can turn an otherwise lousy or mediocre recording into a professional-sounding recording. This often requires a lot of trial and error, and testing, and intimate knowledge of your equipment's unique noise characteristics. However, there are a few general rules of thumb you can follow to quickly maximize the potential of your recordings:

1. Use the gain on your lowest-noise devices to amplify the sound, and minimize the gain on devices with the noisiest pre-amps.
2. Set the gain levels on each pre-recorder device in your chain to provide the cleanest, powerful signal you can to the recorder itself, and keep the input gain low on your recording device.

The discussion on bit-depth and rampling rate above explained how digital audio is sampled, and the impact that different bit depths and sampling rates have on the final quality of your recordings. Now it is time to put that academic knowledge into practice.

If you use high-quality microphones, a high-quality microphone pre-amp, quality cables, and perhaps a high-quality mixer (with high-quality pre-amps) to provide your recorder with a strong, clean audio signal, the last thing you want to do is record it at a bit depth and sampling rate that fails to accurately represent that pristine signal. It makes no sense to spend all that time and money preparing every device in your audio chain from microphone to recorder, to produce a nice clean signal, just so you can record it at such a lousy digital resolution that it will sound terrible when it's played back.

And this is where the CD standard becomes truly useful: If you have done everything right before the signal reaches your recorder, and you record a truly pristine audio signal at the CD standard of 16-bits/44KHz, you're going to love the results! There won't be any need to even consider 24-bits, or 48/96KHz. The CD standard will provide you with an accurate digital representation of that clean signal that you worked so hard to send to the recorder.

However, if you fail to provide a truly clean and noise-free signal, then again, it doesn't matter whether you record at 16 or 24 bits, 44.1 or 96KHz, because in any case you will be digitizing a lousy audio signal. And that means you will only be able to play back a lousy audio signal.

Record at 24 bits and 48/96KHz if you want to. It doesn't hurt to raise the bar. As a matter of fact, 24bits/44.1Khz is probably the ideal compromise between audio fidelity and the size of your digital audio files (since 48KHz and 96Khz sampling rates create much larger files). Still, in most cases, for all except the very most discerning ears--which are usually canine, rather than human--everyone should be happy with your recording at 16-bits/44.1KHz. Best of all, since this is the CD standard, you can burn it directly onto compact disk for distribution.

After you've finally recorded your audio, you need to convert it into some kind of medium that you and other people can listen to. In most cases, the medium will be a cassette tape, a compact disk, or some kind of compressed digital audio file (e.g., mp3, wma).

If your audio will be distributed on cassette tape, you merely need to play the digital audio you've recorded in real time, and record it onto the tape. No file conversion is necessary.

If your audio will be distributed on compact disk, you may or may not need to convert your audio to the CD 16-bit/44.1KHz standard first. There are several codecs and algorithms for doing this conversion, some of which are better than others. Try several of them out for yourself and choose the one that sounds best to you. Once you've converted your audio to 16-bits/44.1KHz, you can burn it directly to an audio CD.

If your audio will be distributed as a compressed digital audio file, the same situation exists as with converting to the CD standard. There are many different codecs you can use to compress your audio into an .mp3 or .wma file for distribution. However, in this case, there is one other important factor to consider: compressed audio necessarily suffers from degredation in quality, so it is important to set your compression parameters to achieve the balance you desire between audio quality and file size.

For example, if you compress a 16-bit/44.1KHz recording into .mp3 format at 320kps--which is pretty high-quality for an .mp3--you will end up with a fairly large file, yet the audio quality may not be noticeably less than the original. However, if you compress to 64kbps in order to create a very small .mp3 file, you will notice a serious degradation in audio quality. These are the trade-offs that come with digital audio, so it's a good idea to familiarize yourself with various audio conversion codecs and parameters, so you'll be able to determine the best type of audio compression for your purposes.