88.2kHz?

[jh]

Quote:I think you have enlightened half of us, and thoroughly confused the other half of us I hope I enlightened the other half ...or then I just confused everybody all over again...

[Casey H]

OK, since I confused half the world, here are some simple exaggerated examples that might clarify sampling rate and bit resolution.First, sampling rate:Let’s say, for argument’s sake you have an analog audio signal that:(1) At time = 10 milliseconds… Amplitude = 100 millivolts (mv)(2) At time = 15 milliseconds… Amplitude = 200 millivolts(3) At time = 20 milliseconds… Amplitude = 100 millivoltsOK, now you sample at only 100 Hz. That means you take a sample every 10 milliseconds starting with (1) above. You would see the 100 mv at (1), completely miss the 200 mv at (2)!!!, and get the 100 mv again at (3). So you see, if the sample rate is too low, information is lost.Now, bit resolution:Pretend you have only 3 bits of resolution. That means the only possible values you can convert your audio signal are: 000, 001, 010, 011, 100, 101, 110, or 111 (8 values). Let’s say your input audio signal ranges from 100 mv to 800 mv. That might mean:100 mv = 000 (binary)200 mv = 001300 mv = 010400 mv = 011500 mv = 100600 mv = 101700 mv = 110800 mv = 111Look at the above table and ask yourself, “What happens if the input is 143 mv? 678 mv? If these get rounded off to the nearest step, 300 mv looks the same as 349 mv (binary 010), 755 mv looks the same as 799 mv (binary 111), etc. Not very effective as you can see… So you need a reasonable number of bits to produce good audio.I hope now only 1/4 of the population is confused... Casey

[hummingbird]

What happens if the input is 143 mv? 678 mv? If these get rounded off to the nearest step, 300 mv looks the same as 349 mv (binary 010), 755 mv looks the same as 799 mv (binary 111).... coffee, I need coffee my head hurts

[Casey H]

Quote:What happens if the input is 143 mv? 678 mv? If these get rounded off to the nearest step, 300 mv looks the same as 349 mv (binary 010), 755 mv looks the same as 799 mv (binary 111).... coffee, I need coffee my head hurts Aw, sorry... BTW, there is a great song by the band Powervibe called "Ones and Zeros"... Perfect for a nerd like me...

[rcase]

Now wait a durn minute, Casey..You don't get to be artistically gifted AND be an engineering/math whiz too I confess, I'm one of those types who- when stumbling across an article in EM or EQ etc containing this level of depth- skips to the final section where the author will take pity on the simple (lazy? ) among us and just tell us what we should use!Rob

[matto]

Poor Elser...you ask one harmless little question...and

[Casey H]

Quote:Poor Elser...you ask one harmless little question...and Ask an engineer what time it is and he will tell you how his watch works...

[sgs4u]

It is nice to see this other fabulous facet of the Casemeister. Who woulda ever thought he was that smart! Why is the sky blue?Quote:Quote:Poor Elser...you ask one harmless little question...and Ask an engineer what time it is and he will tell you how his watch works...

[ddusty]

Quote: Why is the sky blue?Well Steve, everyone knows it's because blue has the shorter wavelength, and it is refracted by the...Oh wait....Nevermind

[ernstinen]

Quote:For 88.2KHz, the samples are taken every .00001134 seconds (e.g. twice as often).At 32 bit resolution each sample step is worth 2.33 x 10^-10 of MAX.There is a point at which faster sample rates and/or increased step resolution provides no additional value to the human ear but make file sizes larger. That's for the audio engineers to tell us.Anyway, I hope I did all the math right and explained it all correctly.Awesome, Casey! O.K., I'm an audio engineer but not techhead by any means. Maybe Batzdorf will chime in!First off, I agree that higher bitrates increase audio clarity better than higher sample rates. I've done blind tests, and that's what my ears tell me. Use the "key ring" test with a good microphone and you'll hear the difference.That said, here I go again with frequency response info! As far as I know, a sample rate is divided by 2 to obtain the highest frequency response. So 44.1k divided by 2 = 22.05 kHz. 48k = 24 kHz etc. (Since analog tape goes well over 30 kHz, these lower sample rates don't capture the high highs in comparison --- but that's another subject! ).NOW, when you get up to 88.2, that's another story. You're talking a frequency response up to 44 kHz! BUT, in my listening tests, even though going from 16 bit to 24 bit is obvious, going from 48 to 88.2 (or 96) isn't as discernable. It's a LITTLE clearer to my ears, but maybe not worth it to most engineers. I mix and master at 24/96 and go down to 16/44.1 as a last step when I make CDs.Casey, please explain why recording at 88.2 is preferable to 96 mathematically when you end up at 44.1 anyway! I know it's divisible by 2, but why is that better for the sound? Inquiring minds want to know!Ern