Before you use 3xosc you should be aware of this

[Rainbow_Rage]

Edit: Turns out I'm an idiot and forgot that everything get's anti-aliased in rendering which makes this a moot point. I just saw an opportunity to bash 3xosc and jumped at it. Still, it's a good reason to render your tracks instead of recording them (I've heard of it happening).

So I've graduated beyond using 3xosc a while ago, but I just recently discovered yet another reason to ditch it that you guys who still use it should know about.

It brings in some higher frequencies that it shouldn't. To see what I mean, put on a pure sine wave, it should only have the fundamental right, well there's some higher frequencies in there too. Still not seeing it? Set the freq to about 50hz and boost your highs. You'll hear a high pitched screech which is not something you want out of a low pitch sine wave.

It does the same thing with the triangle wave (I know its supposed to have a few harmonics, but not like this). I would presume the other waveforms as well but as these have much richer harmonics it's hard to tell.

[Lavender_Harmony]

I'd recommend these as an alternative.

Noisemaker
Elek7ro
BassLine
U No 62

[Captain Ironhelm]

Run a low pass filter, problem solved.

[Rainbow_Rage]

Run a low pass filter, problem solved.

use a better synth, problem solved better.

[Thyrai]

And if you're looking to spend some money

Sylenth1
Massive
V-Station

But yeah, when I used FL I remember getting some weird harmonics from 3xOSC.

[NightmareSnake]

I don't even use 3xOsc except for when I want an 8-bit square wave sound, and even then I use Massive for that half the time.

(or I do it on FamiTracker)

[Captain Ironhelm]

I got into my Massive and did a single straight sine at a very low frequency (think 60-80). I lowered out the lower frequencies and boosted the higher frequencies, and here's what I got:

https://dl.dropbox.com/u/80581976/Samples/MassiveSineWave.wav

Indeed it took more boosting than the 3xosc (Thanks for the heads up on this topic! Been fun messing around with stuff. And 3xOsc is messier than Massive, just to make that clear), but those overtones still snuck through. The same has occurred with other synths. I think it would be almost impossible to get a synth that doesn't do this, however small the degree that it does.

Yay for me giving free worthless information!

EDIT: I boosted it by 120 db, and be aware I did compress it like crazy

[Lavender_Harmony]

Absynth is much cleaner:

Absynth sine high end harmonics Don't worry, its not very loud.

These are the highest peaks, they are not constant, they pulsate softly, so this should be a guide as to where to low pass, about 2000Hz would do, just don't do the hi-pass I have here



As you can see though, it took 96dB of boosting to get these harmonics even barely audible. Yay additive synthesis. I use Absynth for all my sines.

[ChocolateChicken]

I've found that with most synthesizers, the harmonics that are generated with pure sine waves actually come from the filter itself. Turning every filter off in the synthesizer will usually get rid of these harmonics from the sine. This also eliminates the need to use a low pass filter to snub out those harmonics, as they will no longer be a problem.

[Rainbow_Rage]

I got into my Massive and did a single straight sine at a very low frequency (think 60-80). I lowered out the lower frequencies and boosted the higher frequencies, and here's what I got:

https://dl.dropbox.com/u/80581976/Samples/MassiveSineWave.wav

Indeed it took more boosting than the 3xosc (Thanks for the heads up on this topic! Been fun messing around with stuff. And 3xOsc is messier than Massive, just to make that clear), but those overtones still snuck through. The same has occurred with other synths. I think it would be almost impossible to get a synth that doesn't do this, however small the degree that it does.

Yay for me giving free worthless information!

Yes, no synth is perfect, but 3xosc is significantly worse than any other synth I've tried. Most synths need hell of a boost before those higher frequencies show up while with 3xosc it is noticeable flat and only needs a little boost in the highs before it's really obvious.

[Seven]

Wellp, at least this explain my failed attempts to transfer a couple patches from FM8 to 3xOsc. (FM8 was hogging CPU for patches that really didn't need FM8.)

[Peak Freak]

I am not sure about the details of how 3xOSC works, but it could be simple aliasing errors causing these high frequencies :/

[XXDarkShadow79XX]

As it turns out, there's this weird EQ thing and it can actually take those frequencies out.

[Lavender_Harmony]

Could someone do an allpass boost of the 3xOSC sine and post up a frequency graph? Cheers

[bartekko]

In my, humble, honest opinion, the biggest problem with many sofware synths lies not within 50Hz sine waves, but within high-frequency sounds that have loud harmonics going above nyquist frequency of your sample rate.
For example, in Absynth5 and FM8 it's very easy to make patches that do that.

Let me tell you about http://en.wikipedia.org/wiki/Nyquist–Shannon_sampling_theorem
it says that: If a function x(t) contains no frequencies higher than B hertz, it is completely determined by giving its ordinates at a series of points spaced 1/(2B) seconds apart.
To translate into english, If your signal contains no frequencies higher than half the sampling rate, it can be played back without any frequencies going crazy on you

As an example, I'm giving this .wav file:
https://dl.dropbox.com/u/43557387/882L882441L441M.wav
Warning: the provided file is rendered in 88200 Hz, so if you play it back at 44100 Hz, you'll hear something entirely different for the first half of the file.

These are four renders of a C-major scale played at a very high pitch, with an FM8 patch with a lot of high frequency content.
1.) A render at 88200 Hz with a Lowpass filter @ 20000Hz cutoff (Allpass Filter)
2.) A render at 88200 Hz
3.) A render at 44100 Hz with a Lowpass filter @ 20000Hz cutoff (Allpass Filter)
4.) A render at 44100 Hz

First three parts sound pretty much the same, with mostly clean sound, and no weird frequency shifter-like tones, but the fourth one sounds really horrible and digital, doesn't really sound like the C major scale.
The reason for this, is that the last part has a lot of frequencies above the nyquist frequency, which is 22050Hz for the 44100Hz sampling rate, and those frequencies are rendered as different ones, as the image shows:

Black dots are samples, Red wave is the input wave higher than nyquist freq, black wave is the output, at frequency of f=f(sampling)-f(signal)

Play the file at 44100 Hz (Half Speed) (You may use Audacity)
The first and the third ones sound similar, but the second one sounds more briliant, because it has all the frequencies that distorted the fourth signal in the fourth one rendered correctly



So the point is, if there was an idiot engineer who forgot to make the synthesizer render in a higher sampling rate and then lowpass the output just before the aliasing to output sampling rate, all synths will sound horrible.


Thank you for listening, and Goodbye

[5COPY]

bartekko's post trulely have enlighten me with knowledge on this topic. Digital Synths will always be digital synths after all

[the4thImpulse]

(Everyone needs to know this so read it in his post)

One thing I would like to add, although it will get more offtopic.

All this sample rate stuff applies to EVERYTHING digital. Any sound that gets recorded on to a computer (or other digital medium) goes though this process when being recorded, and when played back from the computer on monitors or whatever other transducer you use. Using sample rates around 44.1 kHz when recording vocals or guitars and every other instrument with anything in the high end is not good enough. It doesn't take a trained ear to hear the difference between 48 and 96 kHz so always record and mix your music at the highest possible bit rate your computer will allow.

Higher sample rates will mean more processing power is required aswell as more hard drive space to hold the samples (sounds/recordings).

[prettiestPony]

Might I suggest that one of you make a new thread about this topic to inform the general public? I do think that's important but oft-overlooked information, and it might be nice for others to learn about it without having to wander into this 3xOsc thread by chance.

[StevenAD]

As it turns out, there's this weird EQ thing and it can actually take those frequencies out.

That's what I do, although I'm finding I'm using 3xOsc less and less as I discover more things about other synths.

[GumsOfGabby]

As it turns out, there's this weird EQ thing and it can actually take those frequencies out.

That's what I do, although I'm finding I'm using 3xOsc less and less as I discover more things about other synths.

Try taking those extra frequencies out on a saw tooth waveform.

[DJColeTPony]

Its time for the bucking wall of SCIENCE!

Let me tell you about that: http://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampling_theorem

If a function x(t) contains no frequencies higher than B hertz, it is completely determined by giving its ordinates at a series of points spaced 1/(2B) seconds apart.

For example. Bartekko gives us a wave file and says:

As an example, I'm giving this .wav file:
https://dl.dropbox.com/u/43557387/882L882441L441M.wav
Warning: the provided file is rendered in 88200 Hz, so if you play it back at 44100 Hz, you'll hear something entirely different for the first half of the file.

These are four renders of a C-major scale played at a very high pitch, with an FM8 patch with a lot of high frequency content.
1.) A render at 88200 Hz with a Lowpass filter @ 20000Hz cutoff (Allpass Filter)
2.) A render at 88200 Hz
3.) A render at 44100 Hz with a Lowpass filter @ 20000Hz cutoff (Allpass Filter)
4.) A render at 44100 Hz

First three parts sound pretty much the same, with mostly clean sound, and no weird frequency shifter-like tones, but the fourth one sounds really horrible and digital, doesn't really sound like the C major scale.
The reason for this, is that the last part has a lot of frequencies above the nyquist frequency, which is 22050Hz for the 44100Hz sampling rate, and those frequencies are rendered as different ones, as the image shows:

Black dots are samples, Red wave is the input wave higher than nyquist freq, black wave is the output, at frequency of f=f(sampling)-f(signal)

Play the file at 44100 Hz (Half Speed) (You may use Audacity)
The first and the third ones sound similar, but the second one sounds more briliant, because it has all the frequencies that distorted the fourth signal in the fourth one rendered correctly

Look at the waveform-thingy-a-blob.

The samples in the waveform-thingy-a-blob. The samples will be identical. With a waveform-thingy-a-blob and the magic with 3xOsc. You can make Random Sounds will responded on the WAVEFORM-THINGY-A-BLOB. Now read from the start!

The more you know. The better you make music!
Now thanks for listening. Ciao for now.

WE DID IT. WE HAS A GREAT TEAMWORK

[cloudshovit]

I still like using 3xOSC. can't any CPU hit from that thing. And you can drag and use whatever waveform you like, the only thing I don't like about it is the preset management since its not a VST

[cloudshovit]

Wellp, at least this explain my failed attempts to transfer a couple patches from FM8 to 3xOsc. (FM8 was hogging CPU for patches that really didn't need FM8.)

errrr I don't think thats possible tho. FM8 is an FM synth and 3xOSC is whatever...

mmm never had cpu hogging problems with FM8. even when I'm just using it on my laptop.

[Nine Volt]

NEEEEEEEEECRROOOOOOOOOOOOOOOOOOO!!!!!
:3

[bartekko]

welcome to 10's. please leave your 00's in the crusher