Sample Rate Confusion

Sorry, to provide more insight: your software will poll the audio file 48,000 times a second even though it was recorded only with information every 44,100 times a second.

The question then, what does it do with the “extra” polls that the file doesn’t have information for? I actually, to be totally honest, don’t really know. I imagine it would interpolate the frequency, or simply use the same one as it used last measurement.

Hi guys,

Sample rate indeed means how many times a second of sound is chopped into for measuring. Some clever guy decided that in order to be able to digitize sound waves, the sampling rate needed to be at least twice the frequency range. With human ears hearing up to 20KHz, the minimum rate needed to be 40KHz. For a few (mostly technical reasons) for CD’s they decided on 44.1 KHz sampling rate and 16-bit depth (meaning that every one of those 44.100 times a second a measurement is taken, there is 16-bit available to “describe” the level. So, all consumer hardware (CD players) can and will read 44.1/16, and most of the digital processing done will also be geared towards that setting.

On DVD’s sound became sampled at 48KHz, so that became another “standard”. Here too there was mostly “non-musical” reasons for coming up with that sampling rate.

The bit-depth determines the resolution with which you can “describe” the level of every sample. 16-bit means you have about 64.000 steps between 0 and top/bottom of a waveform. 24-bit brings that to 16 million. The major advantage of this is the increased headroom. Hence most recording is now done in 24 (and even 32) bit depth.

Higher rates (up to 96 KHz) are used primarily in sound treatment. The end result then being brought down to (in music’s case) to 44.1. During this process some information is invariably lost, but because less was lost than if you had done all your manipulation (filters, eq and such) on 44.1, the end result is “better” (although perhaps not all that audible to the usual hifi listener and most likely not for you regular club visitor – especially not the ones who, rightfully, wear hearing protection!).

Although rates go up to 196KHz, many frown at that, since it seems to introduce some unwanted artifacts into your sound signal.

What you see in some controllers, is that they will have higher sampling rate/bit depth internally (i.e. where all the processing is done) and have 44.1/16 output.

Quite frankly, giving our usual audience and playing environment of less than optimal acoustics, speaker systems that are tuned towards long-term, high-volume and not towards listening levels hifi quality, high ambient noise floors (the buzz a crowd makes itself) and the presence of many sound-altering hearing protectors, I wouldn’t worry too much about output rate/depth.

The thing to watch for is that to play 48KHz/24 music on a 44.1/16 bit system, means the software/computer has to be busy doing the “re-sampling” (term for convenience sake). So it’s best to have your base material in 44.1/16. This is true for every CD-based track.

The question you ask is, why not do everything at 96KHz and 24-bit then?! First of all, it’s not that by doubling the rate you get double the quality. And as I sketched before, in our line of business and with our gear, I doubt anybody would hear the difference (or care, for that matter!).
The obvious downside to higher sampling rate/bit depth is … the amount of (disk) space needed!

The difference for a 3-minute song between 44.1/16 (30 MB) and 96/24 (99 MB) is a solid factor 3.3 and here we all are trying to get as much music on our Usb-sticks, iPods and such that we even compress it further to MP3/AAC.

In conclusion: if you make your own stuff, try to do at least 48/24 and possibly 96/24 when you are producing or working on recorded material. But always bring it back to a 44.1/16 CD-quality stereo master output that you’ll use (either compressed to MP3/AAC or as pure Wav or lossless). For anything pre-packed 44.1/16 is CD-quality and that is good enough for our purposes, imho.

Hope that helps.

Greetinx.

If the sample is 44.1K then that is the upmost quality you get. If you re-render with your audio interface at 48K then you get the original quality of 44.1K but now it’s at 48K. Like if you made a WAV of a MP3 you’d get almost 10x the file size for what still sounds like a crappy MP3. Interesting note here is that you can run a DAW at 44.1K and compose at lower buffer, because you save on CPU, and then render out at a higher sample rate for quality and mixdown the stems where buffering is not so important.

Marketing, mainly.
Though, it’s not bad to have something that CAN play higher rate stuff in the rare case you have a file like that, is it?

My CDJ’s can’t even play MP3’s not at 44.1 kHz lol.