Number of Samples in a selection?

hmmm....interesting...I don't know the answer, unfortunately. Just curious, but why would you want to know this? I suppose the easiest way to figure it is to multiply the length of the selection (in seconds) by the sample rate (44,100 or 48,000 or 96,000 etc.), but you probably asked in order to avoid doing that...

I don't think it can be easily determined. Even if you did a multiplication of time and length you would have to account for frequency. If its a sine wave yes it could be calculated but music is a complex wave form with multiple frequencies. I could be wrong though.    

I'm surprised to see you say that, John. I'm sure you know very well that sample rate is constant, and samples/second x seconds = samples.

I'm thinking about frequencies and that is not constant. A single frequency would have x number of samples at a given sampling frequency. However if a low frequency has also a high frequency along with it the total number of samples seems to me to vary. Don't you think? If for example a 20 kHz frequency is sampled one  sample is the result at 40 kHz sample frequency if a low frequency is also present in that time frame that low frequency is also sampled and and it should not be just one sample. The idea is if the low frequency is 1 Hz then at a 40 kHz sample frequency it should have a little less then 20,000 samples.

John

I'm surprised to see you say that, John. I'm sure you know very well that sample rate is constant, and samples/second x seconds = samples.

I'm thinking about frequencies and that is not constant. A single frequency would have x number of samples at a given sampling frequency. However if a low frequency has also a high frequency along with it the total number of samples seems to me to vary. Don't you think? If for example a 20 kHz frequency is sampled one  sample is the result at 40 kHz sample frequency if a low frequency is also present in that time frame that low frequency is also sampled and and it should not be just one sample. The idea is if the low frequency is 1 Hz then at a 40 kHz sample frequency it should have a little less then 20,000 samples.

You are apparently thinking in samples/cycle. But sample rates are in samples/second, and are constant. So, yes, a higher frequency signal has fewer samples representing each cycle, but as you know per Nyquist, two samples per cycle are all that is needed to perfectly reconstruct a signal component's frequency, amplitude and phase.

EDIT: Is this a test? Or has John's brain been taken over by an alien life-form from planet Stair-step? 

So what is being represented when we zoom all the way in on a wav in Sonar and see the what looks like steps?

That's SONAR taking the easy way out, and drawing lines between sample points. An actual D/A convertor uses fourier transforms to reproduce an analogue signal from those same data. More sophisticated audio editors use the same algorithms to draw a smooth waveform trace through the sample points, complete with inter-sample peaks.

jinga8


 Just curious, but why would you want to know this? I suppose the easiest way to figure it is to multiply the length of the selection (in seconds) by the sample rate (44,100 or 48,000 or 96,000 etc.), but you probably asked in order to avoid doing that... 

Thanks for the replies. 

  I was investigating latencies of my audio and bounced-vst tracks and I was getting tired of doing the math! 

brundlefly
...nudge values of 24, 12 and 1 samples set up. Then count your nudges to get source and re-recorded tracks aligned. Using 24 and 12 makes it easy to count fractions of milliseconds if preferred. 

great tip...thanks