sampling and depth rate

[radiorickm]

there is a forum open right now in the newbie stuff called VOCAL MODIFICATION. The guy says that he recorded his voice into program X and when he burned it to CD, his voice was an octave lower. When he asked why, the answer given was SAMPLE RATE differences.

I say, no. not even close.

Well, I am not a digital GURU. But here is how i understand this, and hopefully it will help other new people.

First thing: SAMPLING. Sampling is not interacting with or changing. it is taking a picture of what is there.

So, here is how i think this works.

Picture yourself, setting on the bank of a river, watching the water run by.
You decide to analyze the water. So, what do you do?

Well, you take a sample of it, of course. But one sample may not reflect what actually exists. You would probably want to take several samples. If you took one sample a year, think how much things could change in that year. To be more accurate, you would sample much more often; let's say once a day. We changed our sample rate from 1py (thats one per year) to 365py. See the potentional for a little more accurate info here. Now, because you are a super guy, you decide to take a sample once a second. Your sample rate is now 2,073,600py (Thats 2.07Mpy by the way). Once again, all of thes samples are bound to result in a much more accurate picture of what is out there in the river.

But, there is one more question in the equation. THat is, how much water are we analyzing at a time. If there are one hundered thousand gallons of water flowing past each second, and we only take one teaspoon of that water, once again this reflects and extremely small percentage of the available total. If we could take all 100,000 gallons and analyze that, it would be great. unrealestic, but great.

So, we have to come to a happy medium. We could sample all 100,000 gallons, once a year, or we could sample a teaspoon a second.

SO we come to an effective balance, between the sampling rate, and the sampling depths. Lets say for the heck of it that we end up sampling 1 gallon every minute for a year. that is a comprimise, out in the middle that fits our needs. It takes us lets say 10 seconds to gather up one gallon of water, and 50 seconds to analyze it. This is an effective comprimise

We do this in the computer also. It can only analyze so fast. Sure, it is getting faster and faster with each new computer, but still it only goes so fast.

So lets put ENGLISH back in computer-nese.

The sampling rate, is how often the program goes out and looks at the data streaming past it. In most cases, 44kHz means 44,000 times per second. (44, k equals KILO, or the metric prefix for thousand, and Hz is Hertz, or the measure of occurances per second)

The sampling DEPTH is how much information is stored. It stores 96Kb of info each time it looks. If it only stored 2b of info, A teaspoon full, that would be either 1 or 0. That would tell us whether there was water or not in the river, but that is all we would know. It stores 96,000 bytes of info that it took, or a much bigger "sample of the water".

Well, This is how i understand, and can relate the digital storage questions everyone has.

If i am really wrong here, someone please tell me.

happy recording

 

[RawDepth]

Yes, you pretty much got the idea. The bit depth part is not entirely accurate, but it’s close enough to understand how to set things when you record. As long as you are in control of your software, you are well on your way to making decent recordings.

As I understand it, increasing Sample Rate will improve upon the resolution of frequency and increasing Bit Depth will improve upon the resolution of amplitude.

 

[AlChuck]

Interesting analogy, though I find it more confusing than the straightforward explanation.

But the statement about the lowered pitch, to which you said, "I say, no. not even close..." well, it is close, in fact, it's undoubtedly correct.

The point about the answer in the "Vocal Modification" thread was that, if you have an audio data stream that was recorded at 88.2 KHz sampling rate, and played it back at 44.1 kHz, it would play back at half the speed, equivalent to what happens when you play a tape back at half the speed, and the sound is an octave lower.

What I suspect is that the guy first recorded it at 88.2 or 96kHz, and then burned his recording to CD in some boneheaded way that, instead of downsampling for him on the fly, left all the data in there and simply created a CD Audio data stream at 44.1 kHz, twice as long as the original and therefore an octave lower.

 

[noisedude]

Yup, that sounds right. A musical octave is a doubling or halving of a frequency. So it's be like playing a record at half speed.

 

[Casaubon]

but it would have surely been at a much slower speed (the whole song or just his voice?)

 

[AlChuck]

Of course, half the speed it the frequency was halved.

 

[radiorickm]

bottom dollar

hmmm. i would have about bet my bottom dollar on that one!

guess we learn something new every day.

i assumed (i know, i shouldnt do that) that he put the music in the same way, and it should have been dropped too. guess thats not what happend though.


anyway, thanks for educating ME!

 

[Agent47]

Jees, that's a long-winded and confusing analogy!

But an accurate one nevertheless...