Should Subtractor really be aliasing this badly?

[househoppin09]

Despite using Reason for a great many years now, I've generally shied away from Subtractor in favor of Thor and various REs. Recently I've decided to try to remedy that, as I do have fond memories of using it in the early days, and its simplicity masks considerable power. Unfortunately, I've been floored by how bad the aliasing seems to be. Yes, a certain amount of aliasing can lend some pleasant "grit" to a synth, but this is ridiculous--for cripe's sakes, the alias frequencies are loud enough to distract from the fundamentals at times! I've seen a lot of people here defending Subtractor's output quality, claiming that it doesn't have any noteworthy problems with aliasing, etc. My experience is so starkly different from this that I can't help but wonder if there's something wrong with my copy of Reason. Could it be that the naysayers are simply running at higher sample rates that are masking the severe aliasing? I'm running at 44.1, and it's just brutal.


I've attached audio examples. This is just a brand new, blank Reason song, and a single Subtractor on Init Patch. The only alteration I made to the Subtractor's Init Patch settings was to lower the main filter cutoff from 64 to 48, and raise the output level from 100 to 127. No other tweaks. I then played a scale running from A4 up to A5 and back down to A4. We're not talking about unreasonably high-pitched notes here, folks! At 44.1, the aliasing is mind-boggling--there's just no possible way to get anything even close to a normal, unembellished saw, as you can hear in the file marked "44.1". Compare that to the version marked "192", which, as the name indicates, is simply the exact same Reason song exported at 192 with no other changes. Mind-bogglingly, there's still some audible aliasing in the A4-A5 octave, even at 192! However, as you can hear, it's subtle enough that it just lends some nice "grit"/"character" to the sound, so it's not necessarily a problem, per se.


I think it's pretty clear that I'm not misunderstanding or misinterpreting anything that's going on here, but I'm baffled that this severe aliasing isn't mentioned more often here as a glaring limitation of Subtractor's ability to create important, very basic kinds of sounds. What's even more baffling is that I've seen so many people outright denying that this is much of a problem. Combined with the fact that I don't particularly remember being so affected by this issue when I used to use Subtractor back in the day, I'm really wondering whether I'm experiencing some kind of pathological or unusual behavior here. I would greatly appreciate it if some of you could check out these samples and let me know whether this is how it's supposed to be--and, if so, explain how the heck this isn't considered a showstopper.


Somewhat bizarrely, there seems to be a hard limit of 1 MB for attachments, so I've uploaded a zip of the two audio files here: https://www.sendspace.com/file/xwi1bu

(Sorry about the crappy download site, but it should be safe as long as you just stick to the blue "download" button and strictly ignore any popups. I don't have a Dropbox or Google Drive set up and couldn't think of any better way to share this on short notice.)

[mataya]

Yeah, I hear a nasty out of tune ghost melody behind yours and it's pretty bizarre.
Not surprised do.

M

[bitley]

No need for audio examples on the most common synth in the rack. Of course it suffers from aliasing but the job of the synth programmer is to work around that. I love the subtractor for what it is, I can do just about anything I need with it very quickly. The end result doesn't suffer from aliasing; in a mix nobody cares. A good sound is a good sound.

[Benedict]

The infamous Virus aliases very badly in certain situations. Don't recall anybody saying it was a bad synth.

Bitley is right. Sub is what it is and it is a very good synth (as you say yourself) and like everything has its character points and it is up to the sound designer to work with that. Many old tech devices are now famous for the very things that could be seen as fatal flaws - TB-303 anyone.

Didn't listen to your example but I would question running Sub's Vol at 127 as he is prone to being pretty loud so you could just be overworking something else without realizing. I was just using Sub today and at 44k I heard no aliasing issues.

[RandomSkratch]

You don't happen to have Low BW enabled do you? Could that be a source of increased aliasing?

[chimp_spanner]

Yeah Subtractor kinda is what it is I guess - it's still a workhorse synth and I use it almost all the time, but if I need extreme detail and super fast modulation rates I just use any of the virtual analogue racks that are out there But I mean even Thor can be pushed to crazy extremes and it holds up pretty well!

[drloop]

Yes the Subtractor suffers from aliasing at higher pitch sounds. But it is still a great synth.
Here´s some Subtractor love, all sounds Subtractor.

[househoppin09]

You don't happen to have Low BW enabled do you? Could that be a source of increased aliasing?

Nope! First thing I thought of. Great minds think alike. It's not enabled.

Didn't listen to your example but I would question running Sub's Vol at 127 as he is prone to being pretty loud so you could just be overworking something else without realizing. I was just using Sub today and at 44k I heard no aliasing issues.

You're right that Subtractor at 127 is loud, but I don't see why that should be an issue here. I'm well short of clipping, and there are no other instruments in the mix to worry about. And I get exactly the same issues at lower volumes too, of course.


Overall, it sounds like people aren't very surprised by what I'm reporting here, so I suppose it's not just me after all. And, to be fair, the kind of intense aliasing I've demonstrated in the example audio files is only an issue when trying to get a fairly narrow range of sounds out of the synth. Basslines and somewhat less high-pitched leads and melodies obviously aren't affected to the same degree, and neither are many of the other waveforms even at higher pitches. So I guess that's why people don't raise more of a stink about this, then? Even so, It's pretty amazing to think that one of the most fundamental devices in Reason, which for many years was its primary workhorse synth, was literally incapable of running normal-sounding saw notes in the A4-A5 range! If anyone else wants to check out those audio samples and share any thoughts on this, I'm still interested.

[RandomSkratch]

Yes the Subtractor suffers from aliasing at higher pitch sounds. But it is still a great synth.
Here´s some Subtractor love, all sounds Subtractor.

Ohh that's nice

[lowpryo]

I'm with you here. I stay away from Subtractor for that reason. even when I make sounds that don't have audible aliasing, there's just a psychological block that gives me an uneasy feeling, like they're still dirtying up the mix. I'd love if they could add oversampling or some other improvements to the stock devices. computers are much better than they were 10-15 years ago. we should have the option to utilize it!

[Galaxy]

Envelopes are better in subtractor compared to Thor without mod buss trickery.

[fieldframe]

Yes the Subtractor suffers from aliasing at higher pitch sounds. But it is still a great synth.
Here´s some Subtractor love, all sounds Subtractor.

Ohh that's nice

Yeah, this is some quality stuff. Thanks for the reminder of what a beast the old Sub can be in the right hands!

I usually keep Subtractor on a fairly short HP/LP leash. But it is funny to think about how aliasing has historically been an integral part of digital synths – according to one analysis I've read, the Roland JP-8080 sitting right here on my desk actually uses aliasing intentionally as part of its signature supersaw sound.

[RandomSkratch]

Yes the Subtractor suffers from aliasing at higher pitch sounds. But it is still a great synth.
Here´s some Subtractor love, all sounds Subtractor.

BTW did you compose this or is it someone else? I would love to look at those patches and see the composition.

[Benedict]

That does have some really impressive sounds that lever the Subtractor sound very well. I bet people might mistake that for a Virus or Nord (the guy who made Sub).

Very different sound but this album of mine is covered in Subtractor (and NN-XT often samples of Sub) as that was the only synth in Reason 2.5. While the sound was different from other things I had been using (incl. hardware) it definitely doesn't suffer from aliasing in any bad way - even if it is there. I deliberately decided with this record to accept the "Reason sound" and work with it. Was a watershed for me as finally, I learned to stop fighting my tools (myself) and embrace what was on offer.

[drloop]

BTW did you compose this or is it someone else? I would love to look at those patches and see the composition.

Yes I made this short demo. I make short demos when I have done some sounds in Reason.
I am actually working on a refill with some of my sounds, it will be free of charge so justsit back and wait...

[miscend]

Subtractor sounds better at 96k. Most newer synths are using over sampling to avoid aliasing.

[bpmorton]

I'm running at 44.1 as well and the aliasing isn't as bad on my equipment as it is in the provided example. Strange. I use probably the cheapest converter in the world (old version of M-Audio Mobile Pre USB.)

[ScuzzyEye]

I'm running at 44.1 as well and the aliasing isn't as bad on my equipment as it is in the provided example. Strange. I use probably the cheapest converter in the world (old version of M-Audio Mobile Pre USB.)

The aliasing comes from Subtractor, not your interface. The signal is aliased in the way the waveform is turned from a continuous function into discrete samples. Once this aliasing occurs (as the audio is rendered to the Subtractor outputs), there's nothing that can be done to remove it.

[bpmorton]

I'm running at 44.1 as well and the aliasing isn't as bad on my equipment as it is in the provided example. Strange. I use probably the cheapest converter in the world (old version of M-Audio Mobile Pre USB.)

The aliasing comes from Subtractor, not your interface. The signal is aliased in the way the waveform is turned from a continuous function into discrete samples. Once this aliasing occurs (as the audio is rendered to the Subtractor outputs), there's nothing that can be done to remove it.

but why is the aliasing worse on his example? Not saying I don't get some aliasing but it doesn't seem to be as loud as in the example. Are you saying that once the output is rendered as an audio file the aliasing will show up worse?

[ScuzzyEye]

but why is the aliasing worse on his example? Not saying I don't get some aliasing but it doesn't seem to be as loud as in the example. Are you saying that once the output is rendered as an audio file the aliasing will show up worse?

No, rendering audio won't change it. The only thing that causes aliasing is trying to produce frequencies that are above the Nyquist limit. So either the sounds he was using have more, higher harmonics, or the notes themselves were higher than what you're trying.

[selig]

but why is the aliasing worse on his example? Not saying I don't get some aliasing but it doesn't seem to be as loud as in the example. Are you saying that once the output is rendered as an audio file the aliasing will show up worse?

No, rendering audio won't change it. The only thing that causes aliasing is trying to produce frequencies that are above the Nyquist limit. So either the sounds he was using have more, higher harmonics, or the notes themselves were higher than what you're trying.

This ^^ is important - if you don't send any frequencies above Nyquist to the output of Subtractor, there can be no aliasing.

So all darker patches will likely be alias-free. Meaning, unless you run the filter wide open and play notes up high (and use a harmonic-rich waveform such a s saw or square), you'll not hear the aliasing.

In fact, when testing for aliasing it's often difficult to create an audio source that will expose it. I usually use a saw wave with a slow pitch LFO so I can listen for the tell-tail "pitches going in opposite directions" clue (also easy to see on a spectrum display. Without the LFO it's not always easy for me to hear aliasing.

And finally, like distortion before it, aliasing itself has become an 'effect' that is found useful by many. So, it's not always a bad thing to alias…


Sent from my iPad using Tapatalk

[bpmorton]

He's using the default "reset device" patch. So if I reset my device I should hear the same amount of aliasing. I do not. Why is that?

[selig]

He's using the default "reset device" patch. So if I reset my device I should hear the same amount of aliasing. I do not. Why is that?

Different sample rates?

[bpmorton]

He's using the default "reset device" patch. So if I reset my device I should hear the same amount of aliasing. I do not. Why is that?

Different sample rates?

He said 44.1 which is what I'm running at. I'm also running at 16 bit not 24 bit. Maybe that has something to do with it???

[househoppin09]

Hey bpmorton, thanks for chiming in. This is very interesting. I was running at 44.1/16-bit, just like you, when I made my examples (not that the bit depth should really make a difference here, anyway). Have you followed exactly the same steps that I did when configuring the Subtractor? Start with Init Patch, filter cutoff down to 48, output level up to 127, play notes between A4 and A5. (Go by ear to make sure you're in the same octave as my examples, don't just rely on a readout or something.) Are you saying you followed all of those steps exactly and you're not getting the same results as me? If so, could you please export a quick .wav example (at 44.1/16-bit) and post it? Oh, and--this is important!--please make sure you don't have any other devices anywhere, like channel inserts, master inserts or whatever. And no features enabled on the main mixer, like compression or EQ or anything. If your default Reason song has anything at all modifying the signal, we won't get a proper comparison. Thanks.