Get a FREE fully-working copy of Steganos Security Suite 7 (worth £49.99) when you sign up to the Practical PC Newsletter. Do you need a 24-bit card?
In the first of an occasional series of recording hints and tips, Ian Waugh helps you see whether you need a 24-bit digital audio card.
If you keep half an ear to the ground about what's new in the computer music field, you'll know that a many sound card manufacturers are releasing 24-bit and even 32-bit digital audio cards. Maybe you're wondering if it's time to upgrade from your 16-bit jobbie, or if you're looking for your first serious digital audio card, you may be wondering if 24-bit is really much better than 16-bit. Let's see.
First we have to know what the 'bits' are about. There are two sound card specs to look at - the sample rate and sample resolution (this is the 'bits' bit!).
Sample rate
The sample rate is simply how many times per second a sound is sampled. The 'standard' rate is 44,100 times per second (usually referred to as 44.1kHz) and this is the rate used by audio CDs.
The more samples you take each second, the more accurate the digital representation of the sound will be and many systems can use sample rates of 48kHz and 96kHz.
So why not always use the higher rates? One reason is that the higher rates require more processing power, and the file sizes are larger ('cause they contain more data) which requires more disk space.
However, mathematician Harry Nyquist discovered that in order to sample a sound accurately it need only be sampled at twice its frequency. The range of human hearing runs from around 15Hz to 20kHz so in order to sample the highest sounds we can hear - 20kHz - we only need a sample rate of 40kHz. So 44.1kHz ought to capture the full range of human hearing.
Sample resolution
The sample resolution is the measuring scale used when taking a sample. The larger the measuring scale (that is, the more numbers we use to measure the sample), the more accurate the result. 16 bits use 2^16 or 65536 numbers. Audio CDs use 16 bits.
Compare this with 8-bit resolution which only has 256 (2^8) different values. If a sample had a real-world value of 80.75 in an 8-bit system, it would be assigned a value of 81 which is close but it's not accurate. A 16-bit system, on the other hand provides a much finer measuring scale and would convert this to (grabs calculator) about 20672 which is virtually spot on.
Don't worry about the maths - the main principle to recognise is that the more bits you use the more accurate the digital sample will be.
Dynamic range
Now, a 24-bit system has an even finer measuring scale (2^24) and here we can introduce dynamic range. This is simply the difference between the quietest and loudest parts of a recording and it's measured in decibels or dB.
The full dynamic range of human hearing is about 140dB. 16 bits offers a maximum dynamic range of 96dB which is actually pretty close - remember the crystal clarity when you heard your first audio CD?
A 24-bit sample resolution offers a dynamic range of 144dB which is more than most people can discern and a 32-bit system provides a stonking 192dB range which is more sensitive than Bugs Bunny's ears! So systems use more bits than 16 in order to capture a greater range of sound.
Okay, so although 24-bit systems can produce higher quality audio, 16-bit is still pretty good - it's good enough for CDs, after all - so why use 24 bits?
Headroom
In theory, that's a good question. However, in practise, you can rarely exploit the full range of a 16-bit system because you need to leave a little headroom to prevent a recording from clipping. In a real-world situation you may end up using only 15 or possibly 14 or 13 of your 16 bits reducing the dynamic range to maybe 78dB.
That's still shed loads better than the average 48dB of a tape recording but if you want the best audio quality you can get, then a 24-bit system allows you to 'leave' more bits for headroom and still provides a higher-quality output than 16 bits.
The downside, of course, is that 24-bit recording requires more processing power and hard disk storage space but in these days of ever-faster computers and ever-larger hard disks that should not be a major problem.
Downsampling
One question which might have occurred to you is - is it still worth using a 24-bit system when the recording has to be downsampled to 16 bits to put it on an audio CD?
The answer is, yes. Because you are working with more bits in the first place, the overall accuracy will be greater and when the time comes to downsample to 16 bits, the result will be higher quality than a project which was created using 16 bits from the off.
Having said that, if you're quite happy with your 16-bit card you don't have to rush out and change it. However, if you are shopping for a new card you'll find that most manufacturers are moving on to 24-bit production so providing your computer is up to speed it's probably worthwhile looking at a 24-bit card.
But don't forget that in order to appreciate higher-quality sound, you need a good audio system. Playing 24-bit audio through ten quid computer speakers will offer no benefit at all!
