Bit depth. Bit rate. Dynamic range. These are all considerations you should become used to pondering when you undertake a multi-track project or a simple edit. I’m not going to get terribly scientific, but I am going to give you a general look at what’s under the hood when you are working with digital audio files.

Sample Rate

Explained in its simplest form, sample rate refers to how many snapshots of audio are taken per second. It’s expressed as hertz. Audio on CD, for example, has a sample rate of 44,100 hertz. These rates are sometimes simplified to kilohertz: 44.1 kHz in our CD audio example. (It’s important to note that, strictly speaking, sample rate does not mean the same thing as “bit rate”, which is explained below.)

In addition to describing the number of samples captured per second, the sample rate also determines the frequency range of the given material. Halving the sample rate will yield the maximum frequency you could theoretically represent in that file. Using our CD example, 44,100 Hz divided by 2 yields 22,050 Hz. Conveniently, 22,050 Hz is just above the accepted range of human hearing (around 20 kHz).

While 44.1 kHz is most common in everyday applications, others are used frequently. 48 kHz is used for audio on standard definition DVD’s, digital television, and even DAT. Blu-Ray (and old HD-DVD) discs use 96 kHz audio.

Generally speaking, the sample rate you choose for a project or recording should hinge upon the ultimate use for it. If you’re producing audio for DVD, you’d want to choose 48 kHz. And, unless you have a special application for which a specific bit rate is desired, 44.1 kHz is a good default.

Converting between sample rates is easy in Audition, but the result will differ depending on the “direction” of the conversion. For example, if you down convert from 48 kHz to 44.1 kHz, you will lose some bits of audio data. While it may not be audible or apparent, it’s still technically a generational loss of quality. Conversely, if you up convert audio that was sampled at 11,025 Hz to 44,100 Hz, audio bits will be added to the file to fill in the gaps, but the quality of the signal will not improve to that of material that was originally recorded at 44.1 kHz. A great reason to record at the right sample rate the first time around.

Of course, the higher the sample rate, the larger the resulting file will be.

Channels

Stereo or mono? This one requires little explanation beyond a few key points. First, if you are recording vocals from a single microphone, you can save hard drive space (and later processing time) by recording in mono. The same applies to certain instruments like basses. It won’t hurt to record in stereo, however. In fact, I recommend always recording in stereo. It’s good habit to always have the cleanest copy of anything you record available for latest processing. If you recording something in mono, then realize you needed it in stereo, you’re up a creek. Short of recording it again, you have to dip into theatrics to ‘fake’ the stereo image.

In my opinion, you should work on all projects in stereo. Even if you’re recording a podcast that you know will be a low bit-rate MP3 in mono, still work on it in as high quality as possible (and in stereo). Don’t make the sacrifice of quality until the last possible moment.

Bit Depth

This is a more tricky nuance of digital audio to wrap your head around. Admittedly, it’s complicated. I even find myself having to reference specifications all the time. If you find this confusing, you’re not alone. The math is tricky. So, we won’t worry about it. Let’s talk about what bit depth means to you and how to choose.

How many distinct levels of loudness can your audio project/file represent? That’s bit depth. If you don’t ask any questions, it’s really that simple.

At 8 bits, there are only 256 levels of amplitude (loudness) that can be recorded. At 16 bits, you’ve got just over 65,000. Much better! At 24 bits, you’re up to almost 17 million. Pushing the upper limit at 32 bits, you’re afforded just over 4 billion. If you’re vaguely reminded of images and the settings for your graphics card, pat yourself on the back. The same principle applies: an 8 bit image can only represent 256 colors (crappy). An 8 bit audio file can represent 256 separate amplitude levels (also crappy). Typically, if you record below 16 bits, you’re doing something wrong.

I recommend that, when working in Audition, you always work in 32 bit. Yes, the files are bigger. Yes, there’s a slightly higher demand on your CPU. However, if you’re a plug-in or filter lover, you’re not going to have to worry about the “decay” of quality that takes place from this process. There’s so much more information (bits) to work with, that you can rack up hundreds of plug-ins before the audio quality even begins to suffer theoretically. Not to mention it’s virtually impossible to clip audio in 32 bits. But, that doesn’t mean you shouldn’t keep good levels.

(Technical note: the latest bit depth that’s “realistic” is 24. When you’re using 32 bits in Audition, you’re using 24 bits plus an exponent. This is called floating point and has to do with a mantissa and an the aforementioned exponent. In addition to being extremely boring and hard to explain, it’s something most users can ignore.)

Dynamic Range

What’s the difference between the quietest and loudest bit of audio that your digital audio can represent? That’s dynamic range, and it’s directly related to bit depth. As I explained, a higher bit depth equals more possible amplitude values. There you go. Dynamic range is normally expressed as decibels (dB). The normal range of human hearing is around 96 dB (the dynamic range of a 16 bit file). A 32 bit file, for example, has a theoretical dynamic range of 192 dB. But your ears begin to hurt extremely bad at 120 dB. So why the need for such a high dynamic range?

You want that extra data, trust me, even if you’re never going to push out 192 dB (that would kill you, for what it’s worth). Not to sound like a broken record, but there are more bits to work with. It’s more precise. I’m back to my philosophy of “always work with the highest quality you can”.

Bit Rate

This is a data rate derived from the first three topics I covered in this blog. Want to calculate how big a file is going to be?

Bit Rate (in bits per second) = Number of Channels * Sample Rate * Bit Depth

A typical audio CD:

2 * 44100 * 16
= 1,411,200 bits per second

However, your computer represents file size in units of “bytes”. The conversion is simple: divide by 8. Then it’s just a matter of unit conversion.

1,411,200 bits per second = 172.266 kilobytes per second
1,411,200 bits per second = 10.094 megabytes per minute
1,411,200 bits per second = 605.621 megabytes per hour

A slightly larger example, using stereo 32 bit audio sampled at 96 kHz:

2 * 96000 * 32
= 6,144,000 bits per second
= 43.945 megabytes per minute

Hint: plug your conversion into Google literally. For example, type in “1,411,200 bits per second to megabytes per hour” to get the result.