Samples, waveforms, octaves and logic

When i create a new single cycle sample, number of samples 168, which makes a C+6, so i finetune to -7 to get a pretty spot on C.

So far so good, i have drawn a one cycle sine sampe and C4 is at 261.67Hz according to my tuner (GTune).

Now if i draw two identical (or as close as a could make it) cycles within the same number of samples, then i should get a tone that is one octave higher. However when i look at the tuner it still says 261.67Hz.

So why is this so? What am i missing here?

Does it sound one octave higher? What peaks are shown in spectrum analyser?

Drawing by hand will “never” yield the same. Don’t underestimate the resolution of audio, unless you use 6 bit waves at a very low sampling rate - or so. Maybe that tuner plugin is very sensitive in detecting the lowest frequency, and as long as you don’t very_much_exactly match the two cycles each into exactly the half of the space there will be a “lowest frequency” in the sample at the “root” frequency of the whole wavecycle. I guess it works by autocorrelation which will always give the strongest result for 1:1 identical cycles which will be the whole hand drawn thing repeating as opposed to the 2 cycles just matching a lot more poorly. Maybe mixing in noise or other dirt (not to the wavecycle sample, but layered at another rate) or effecting the sound to dirten it up this way could break this “easy match” in favour of the new, stronger fundamental? So one octave lower if you doubled the cycles in dirty fashion, plus 2 octaves lower plus another harmonic at about -5 semitones if you use 4 unidentical cycles, and so on. How strong these harmonics are depends on how the cycles deviate. Additionally you can have different upper harmonics in between the integer harmonics.

Try the following for comparison: draw one cycle (no, you don’t draw sines, or…ugh), then duplicate that sample in the sample list twice. The first duplicate resample i.e. with the “adjust” button, choose a frequency that is half of the current, i.e. if the sample was 44100 choose 22050. then select all, copy, and paste at the very end of the sample. Copy the whole 2x sample. Silence the 2nd duplicate (it should still read 44100 at sample rate). Then “mix paste” with “add” mode. Tada, you now have the same wavecycle at half length and 2x in the original sample - one octave up two cycles. I guess the tuner won’t detect this one “wrong”. But this only works if the original sample size is a multiple of 2 (even) ofc because resampling needs to halve the number of samples.

Actually now i found that my two cycles weren’t as identical as i thought, when i make two absolutely identical cycles it shows an octave higher. I did not realise how sensitive this is, i thought a near match was good enough to count it as two cycles.

Thank you Oops, you seem to know your shit around these thing. :slight_smile: