5

I started learning music production. I created (or generated) a sine wave with the oscillator in FL Studio, but did not add any more sine waves into it (by turning the volume off for the two other oscillators; the built-in synth in FL Studio has 3 oscillators in total). Then, I hold the key (it was the middle A) of my midi keyboard then adjusted EQ to see whether there would be any changes with the sound. Yes, the timbre (or the color -not sure about the terminology) changed.

The interesting thing is and the thing I don't understand is: the process with the electric guitar is logical because it is a 'real' instrument and it creates overtones; when someone adjust the treble-middle-bass knobs on the amp then the amp cut offs some frequencies. But in the aforementioned experiment, there was just a one, single sine wave and there was none overtones if I'm not mistaken. So, which frequencies were cut off by the EQ in the FL Studio? How did this happen?

To show the oscillator and EQ I use in FL I recorded my screen with the audio: https://streamable.com/9b3ifz

4
  • 3
    Are you sure the oscillator was only generating a sine wave? If it's truly a sine wave, EQ should only affect volume, not timbre. (I don't know FL studio, but maybe add a screenshot to the question for those who do.)
    – Theodore
    Oct 20, 2022 at 20:15
  • 3
    There was most likely something either in the oscillator, EQ plug-in, or other processing that is done to all audio in DAWs that created overtones Oct 20, 2022 at 20:43
  • 3
    That there are overtones present is clearly audible in the video. Whether they were added by the video encoding or were actually present in the DAW, I can't say. I think you're assuming that the "sine wave" waveform in the virtual synth oscillator is a perfect sine wave. There's no reason to believe that. More likely is that it's designed to imitate a "sine wave" produced by a real-world analog synth, and real-world systems are never perfect, therefore the sine wave would have to have "unintended" overtones to sound just like a real analog synth sine wave. Oct 21, 2022 at 1:44
  • Thank you so much for the great -all of them-answers.
    – user88063
    Oct 22, 2022 at 0:31

5 Answers 5

5

It's not possible, something went wrong with your experiment.

The audio from your video sounds extremely aliased, and it doesn't match your eq moves, so I will assume that this is entirely a video artifact and not actually what you are hearing. If it is what you're hearing, something's gone terrible wrong between FL studio and your speakers.

The plugin 3xOsc runs in draft mode until you export your project*, and in draft mode, the sine waves generated are a bit less perfect than you might expect. I recreated your experiment, and I have to treble boost about 20dB for the aliasing to be even a little audible on middle A, but if you have brighter speakers and younger ears you might hear the aliasing sooner. Try the experiment again with the Sytrus "default" preset, which generates a much cleaner (although still not perfect) sine wave and compare your results.

In recreating your experiment, I also heard a significant amount of noise generated while moving the eq bands. This is totally an artifact of how the software takes inputs, and you should ignore this.


*Even in render mode, 3x osc is quite aliased, and the sine waves are still not quite sines, though the newest version has an "HQ" mode which is fully anti-aliased.

13
  • What makes you confident the artifacts are aliasing? One could think there's nothing simpler than generating a pure sine wave, what would cause aliasing? Oct 21, 2022 at 0:20
  • 1
    I assume you mean during my recreation of the experiment and not in the video. I think it's aliasing because 1) It sounds just like aliasing. 2) The 3xosc software manual makes 3 mentions of aliasing
    – Edward
    Oct 21, 2022 at 0:59
  • 2
    @ToddWilcox double precision, which is a standard in computing gives at least 10⁻¹⁵ precision. For comparison 16 bit audio CD precision is worse than 10⁻⁵. And that's just about rounding precision, not aliasing. I understand aliasing might be an issue with square or sawtooth waves, which have many high frequency components, but a sine wave has none. There is nothing to alias. Unless there is something else going on. Oct 21, 2022 at 15:00
  • 1
    @user1079505: Quantization noise would occur without aliasing if instantaneous amplitudes were rounded to some value, via process that was independent of sampling rate. Aliasing can occur if rounding errors occur in ways that "almost" repeat at some regular frequency. Even simple amplitude quantization noise can introduce aliasing in sampled audio. As a simple example, suppose one is using a sampling rate of 15,750Hz and is trying to produce a 1000Hz tone, but all voltages are quantized to be either +0.,5 or -0.5 volts [only two possible levels]. Then for the most part...
    – supercat
    Oct 21, 2022 at 20:54
  • 1
    ...the wave should be +0.5 for eight cycles, then -0.5 for eight, etc. except that 250 times per second it would be necessary to replace what would be a run of eight with a run of seven instead. One would thus end up with a sequence of 63 samples of which 32 were +0.5, then a sequence of 63 in which only 31 were +0.5, then a sequence of 63 samples where 32 are were +0.5, etc., thus yielding some an unwanted 250Hz component.
    – supercat
    Oct 21, 2022 at 20:59
8

It is important to note that software synthesizers are not scientific software, instead they try to replicate the sound of "real" synthesizers, and often vintage analogue synthesizers at that. The waves they produce may be labelled "sine" or "square" or "triangle", but that doesn't mean that these waveforms will be mathematically perfect. It is the imperfections in the sound and slight variations in pitch that make a synthesizer sound like a musical instrument and not like a test device in a laboratory. I assume that the sine wave produced by FL Studio is trying hard not to sound like a straightforward sine wave, and that's what you're hearing when you filter it. Besides, the same thing is true for the filter; it may be adding things to the sound instead of only taking things away, in a way that is modelling the behaviour of filters in "real" synthesizers.

6

Looking at the posted video, your "sine wave" isn't a sine wave, see the image below. Those small deviations from the perfect sine represent higher frequency components which are audible.

I don't know why it happens. I hope it's not video compression artifact. Maybe it's some kind of jitter simulated by the virtual instrument, or aliasing, or some artifact added somewhere in FL – in any case this shouldn't happen in a well written system, unless it's intended.

The best for you would be to repeat this experiment by yourself – record the sound and look at the waveform. Perhaps display its frequency spectrum too. Try disabling or manipulating various dials you have available in FL and see if it changes. Try disabling the EQ... look for documentation, though I can't find anything obvious in https://www.image-line.com/fl-studio-learning/fl-studio-online-manual/html/plugins/3x%20Osc.htm except for mention of aliasing (which somehow seems unlike to me for such a big effect).

enter image description here

0

I like the screen recording of your session. What I hear is a pure sine. It sounds sharp and awful, due to lack of other spectral components ("overtones").

You can fiddle whatever you want with the filters, but the only effect is the loudness ("amplitude") of the single frequency sine signal. For example, turning the knob for high pass filter will not change the sound in terms of more or less high tones, because these are not present. Because such a filter also has some effect on other parts of the spectrum, the sine will be somewhat louder or weaker.

Regarding precision, other answers correctly emphasize that your "sine" is not an absolutely pure sine. There is a minute deviation from the pure sine leading to some harmonics. However, I would say that the deviation is very small.

Even if your sine is high quality, there is for example still a quantization effect, because your sine is build from a number of discrete steps, because you are using digital equipment. But this effect is also very minor because of high sample rate and lots of bits in the quantization.

My guess is that you will not be able to perceive the harmonic distortion, and that fiddling with filters will not change the sound perceivably.

6
  • 3
    I rewatched the video on another device to rule out playback issues on my end, and the sine still sounds very not-pure. The deviation (in the video) is not small at all. You should rewatch it on better speakers and compare the sound of the video to a better recording of a sine A.
    – Edward
    Oct 21, 2022 at 18:48
  • @Edward Can FL Studio do a spectrum analysis of the output signal? Before I go through the trouble of hooking up the pc headphone output to my oscilloscope, which has an fft button
    – Roland
    Oct 22, 2022 at 21:24
  • Yes. If you click the oscilloscope at the top of FL studio, it switches to spectrum analysis. You can't adjust anything about this display besides its size though, so I'd recommend a VST such as SPAN in the last effect slot of the master channel.
    – Edward
    Oct 22, 2022 at 22:22
  • @Edward And? How big are the harmonics, in percent of the base frequency sine?
    – Roland
    Oct 24, 2022 at 19:51
  • There are 2 frequency peaks at about -40dB from the sine, roughly, and more which are lower. It might not sound like a lot, but when you have this degree of separation in frequency, it can be very easy to hear.
    – Edward
    Oct 24, 2022 at 22:37
0

Disclaimer: I have never written a software synth app - so this answer is based on theory, not personal and practical experience.

FL, like any digital software, is incapable of producing a pure sine wave which is truly analog - having no discrete intermediate states. That is, a software synth will create a close approximation of a sine wave by creating first an ascending stair-step with very small steps very close together, and a variation of the duration of each step to simulate the rounded shape of a sine. No matter how close in appearance this is to an actual sine wave, it is actually a series of compounded square waves - each state building on or subtracting from the previous state. As such, it will have harmonic content, although much lower level than a square or sawtooth.

It might be possible to get an even closer approximation by using a bandpass filter set to sweep with the tonic - which would reduce the more distant frequencies, but would still allow some through. I wouldn't be surprised if this method is used by at least some software synths.

1
  • Digital audio usually consists of a series of samples at a constant rate. So if I asked some software to generate a sine wave, and that software used a 44,100 Hz sample rate, I'd expect to get a kind of stair-step where all the steps have the same duration (1/44000 s), but the height of the steps varies. And I'd expect it to generate the best possible approximation of a sine wave right off the bat, so applying a bandpass filter wouldn't improve it. Oct 22, 2022 at 14:02

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.