00:00
00:00
Midnights-Ocean
I make music and sounds. I post mostly on newgrounds so people can use them in games and such. My full albums can be found on my home page below.

Age 32, Banana

Music

Old Scool

Ocean

Joined on 3/12/09

Level:
27
Exp Points:
7,718 / 8,090
Exp Rank:
5,040
Vote Power:
6.85 votes
Audio Scouts
1
Rank:
Civilian
Global Rank:
> 100,000
Blams:
0
Saves:
28
B/P Bonus:
0%
Whistle:
Garbage
Medals:
1
Supporter:
3y

Sampling rate: Your tools do not define your talent!

Posted by Midnights-Ocean - November 2nd, 2021


People don’t need to make stuff up or believe urban myths to be confident in their abilities!


So apparently most audio “engineers” on youtube don’t know what sampling rate is @.@;


Sampling rate is not sampling range. Sample rate is the set number of samples taken per second. It stays the same, no matter what frequency of sound you are recording/rendering. Higher sampling rates aren’t for recording higher frequencies (like ones you can’t hear). They are for recording/rendering any audible frequency in HIGHER RESOLUTION. I reiterate, sampling rate is the SPEED at which a computer samples audio. Not the bloody frequency range it samples! The higher the sampling rate, the higher the resolution (sound quality).


The real math is simple: Divide the sampling rate, 44.1kHz (most common/standard), by the frequency of your audio sound, say 60Hz (bass kick), and you get the number of samples per cycle for that particular sound (735 samples per wave cycle in this case). That means, when you record that 60Hz sound, the computer has no more and no less than 735 steps to estimate the shape of the waveform’s fundamental cycle. Like if you laid out 735 square blocks in the shape of a wave. The blocks can be what ever size you want, it's the number of them that's important. Here's an example of a single wave cycle (the shape you would be trying to draw with your blocks): https://1.bp.blogspot.com/-Zer0DTeKD8o/Us0Qc4TIE-I/AAAAAAAABqY/3hoUpAh81To/s1600/sinewave.png So as you can see, the more samples you have per cycle, the better you can fake the curves of real life sound waves. It’s EXACTLY like the number of pixels in a picture. The more pixels/samples, the higher the resolution of the sound/picture.


The higher the frequency of the sound being sampled/rendered, the more cycles that particular sound has per second, thus the fewer samples will be taken of each cycle. THIS IS BECAUSE SAMPLING RATE FUNCTIONS AS A CONSTANT/PER SECOND. A 6Khz sound has 100 times more cycles per second than 60Hz. So out of your 44,100 samples taken per second, only 1/100th of them will be made for one cycle in the 6Khz sound. Compared to the 60Hz sound, sampling a 6Khz sound at 44.1Khz sampling rate, gives you only 7.35 samples per cycle to construct your waveform. Try drawing any kind of convincing curve with only 7 blocks or a coherent picture with 7 x 7 pixels. It can’t be done. What’s worse, MOST SOUND IN THE REAL WORLD IS MADE UP OF COMPLEX WAVEFORMS. These waves include fractal harmonics extending well into the high frequency range. The fractal nature of the physical world is a proven fact. To say sound is the exception, is ignorant. Point is, these real life sounds are even harder for a computer to draw than a sinewave. So at 44.1Khz sampling rate, how many samples would you get if you recorded a 16Khz sound? Answer: 2 or 3. That’s it. 2 or 3 blocks. You literally can do NOTHING but draw a simple square wave with that few. Just 2 giant blocks. One up, one down. There's your wave cycle. A simple square wave.


The application of this math is easily proven by entering your desired frequency and downloading the sound file from onlinetonegenerator.com Sine wave shape should be the easiest to work with and shows the degradation that happens as you sample higher and higher frequencies using 44.1Khz sampling rate. Open the file in audacity. Zoom in to find exactly ONE wave cycle (as is shown in the linked sine waveform above), then count the number of dots. Each dot in audacity is a sample. This is how digital audio works. Because of the math sampling functions under, the resolution of digital audio goes exponentially down the higher the frequency of the sound you are recording/rendering/reproducing. Digital resolution is very high in the bass bands but practically non existent in the higher bands. To get high resolution across the entire audible spectrum, you’d need sampling rates in the megahertz range. Not kilohertz. 8Mhz sampling rate would be adequate. That’s 8,000,000Hz not 44,100Hz. Kind a bit of a difference there. With modern technology, it should be possible but since everyone’s convinced 44.1Khz is fine, I’m not holding my breath for it to change.


This is why engineers who do know how digital audio works, facepalm when people say analog has inferior resolution. Digital has better editing capabilities by far, but not high frequency sound resolution. Modern digital sampling can’t even accurately record a 6Khz complex waveform, let alone a 16Khz one. Analog gear can though. Because it doesn’t need to sample. In analog, recording is a straight shot of electro/mechanical physics. A good reel to reel or vinyl has MORE resolution than any digital audio to date. All those hard edges on the digitally estimated waves also produce distortion (mostly odd order). Just play with a subtractive synth for a minute and listen to how different a square wave or saw tooth sounds compared to a sine wave. Edges matter. Wave shape matters. Accuracy matters. It all adds up.


Most people don’t seem to know this stuff though, usually substituting techno babble they heard on youtube. They have no real reference, so it’s not surprising. Most analog gear is too expensive to buy, too old to function properly, or too hard for noobs to handle. Most what people hear these days, is digital or analog that has been digitized for the internet. Digital decoders have a few tricks to mask the poor high frequency resolution and noise though. However, pretending decoders are perfect (which they aren’t), even if the decoder could completely make up for noise and wave warping, you’d still be missing what the computer didn’t record in the first place. That being all the details and curves of those higher frequency sounds, as they get mangled or left out entirely. There’s no way around it, you loose definition/detail when you digitize.


Some people act like this kind of information somehow insults their talent. To be blunt, if you suck at music or production, you will suck regardless if you are using analog or digital gear. The quality issues brought up in this article have nothing to do with one’s abilities and EVERYTHING to do with the pure mechanics of audio sound quality. Instead of flaming, people should look at what they CAN do. Digital or analog, if you can make something that puts a smile on your face, does it really matter if there are better tools you could have? Is it worth being jealous and petty over it? No. Believe in your self. Make music and be happy!


Boopaboopadoo The end. : P


Tags:

2

Comments

Comments ain't a thing here.