Native Instruments Massive X Synth - Sequel to Massive (Out Now!)

[BlitBit]

Hey guys! So I did the follow up video and I thought before I made it live to my subscribers the experts could watch it and explain to me whats happening. Ive found that the filter compounds the aliasing and the LFOs in Oscillator modes are not band-limited at all! Square waves really break this thing.

Concerning the demo with the LFOs at the end of the video: to me it seems that the LFO waveforms are also implemented as wavetables. However, the LFO wavetable oscillators are implemented in a simpler fashion than the audio wavetable oscillators.

Because "LFO" stands for "low frequency oscillator" I would assume that one entry (a.k.a position) in the LFO table only consists of one single waveform as opposed to a set of progressively bandlimited waveforms (that are selected depending on the target frequency). If such a single waveform is read at the low frequencies that LFOs are usually operated on then this does not cause any bad problems (except that technically the higher frequencies will drop more and more the lower the frequency goes). However, if you switch such an implementation to be used as an audio rate ocscillator then the table is read at audio frequencies and in that case you indeed get severe aliasing as heard in the demo.

My assumption is that using the LFOs as audio rate oscillator is only meant as a fallback or compromise if you already use all the regular oscillators. So I would not even assume that NI "fixes" anything like is assumed in the video because the fix would be turning them into full blown audio rate oscillators which are more costly to implement.

Considering the target audience of the first version of "Massive" (wub wub) I would not be surprised if these artifacts will be used as an effect in the future.

[McLilith]

I'm aware of the debate over whether MX has unwanted aliasing, distortion, etc. I'm not going to comment on the technical details of that now. I just want to say, whatever the technical details may be, I genuinely like this synth and I'm happy to have it!

That's the most important aspect for me.

Like most software, Massive X is going to be updated in the future, likely multiple times. I'll probably like it even better in the future.

[JoeCat]

Just watched Nick Batts' video (SonicState), and if I'm not trusting my ears, I'm certainly trusting his - he's had more synths in his ears than [insert famous person] has had [insert object] in his/her [insert orifice]. And he definitely did not say "listen to this - it's shite." Or whatever my good friends across the pond call it

[EvilDragon]

Bandlimiting LFOs is a pretty stupid thing to do really.

My assumption is that using the LFOs as audio rate oscillator is only meant as a fallback or compromise if you already use all the regular oscillators. So I would not even assume that NI "fixes" anything like is assumed in the video because the fix would be turning them into full blown audio rate oscillators which are more costly to implement.

You are correct.

[HcDoom]

Hey guys! So I did the follow up video and I thought before I made it live to my subscribers the experts could watch it and explain to me whats happening. Ive found that the filter compounds the aliasing and the LFOs in Oscillator modes are not band-limited at all! Square waves really break this thing.

Ok, when you modulate osc with lfo (square wave)...lol...thats some nasty aliasing, sounds very bad. I hope NI will release an update soon with some oversampling options. Back to Hive, Dune 3 and Serum until MX is fixed!

[EvilDragon]

LFO is a LOW frequency oscillator. You can use it as audio rate oscillator but don't expect it to have the same audio quality as the default oscillator. This is pretty normal in other synths that can use audio-rate LFOs, like Waldorf synths. Those LFOs are also not bandlimited (and they shouldn't be).

[BlitBit]

[...]
It's pretty ridiculous to think that NI's devs get the DSP 101 wrong... (talking about the video) IMO, that's not aliasing at all.

Hold on. So you're saying In my follow-up video that the sounds that we are hearing at 1:10 is not audible aliasing? are you kidding me?

I thought it was just some bats echo locating where they are.

[sadowickproduction]

Bandlimiting LFOs is a pretty stupid thing to do really.

My assumption is that using the LFOs as audio rate oscillator is only meant as a fallback or compromise if you already use all the regular oscillators. So I would not even assume that NI "fixes" anything like is assumed in the video because the fix would be turning them into full blown audio rate oscillators which are more costly to implement.

You are correct.

Thats it? No comment on the perceivable aliasing at the beginning?

[vurt]

lfos over nyquist

odd band name

[McLilith]

lfos over nyquist

odd band name

Odd, but I like it!

[Functional]

Cheers for the video, saved me some money

Demo it first. The video does not contain correct information.

STAAAAAP dude...

Dude, you're the one making videos about synth where you try to rationalize your dislike for it on the basis that it has aliasing filters.

So here's a task for you and I bet you're going to have a lot of fun with it: now go search me a synth that doesn't. I'm willing to bet that if I downloaded Serum demo and tried out the filter aliasing, I'd get quite similar results.

This is just you being an ass for no reason. I said yesterday that if you're going to do another one of these without doing any homework on it, you're gonna have a bad time.

We generally care less about filter aliasing because filters aren't used as modulation sources and their aliasing is nearly if not entirely inaudible in pretty much any normal usecase. It's rare, at least in my opinion. Of course feel free to disagree, but I don't think people on the average feed filters with oscs that have C10 or something above that as their fundamental. Do you?

For this very reason, nobody complains about filter aliasing (that is on par with what you're seeing there) that exists in pretty much every other softsynth. In fact, I measured with Omnisphere 2.6 aliasing peaks at -48db! https://gyazo.com/38b3a2f72cf1491aa79c59cc0b550fd3

Want to know what's the best part about this? The fact that your experiments couldn't be replicated in Omnisphere 2.6. Why? Because it doesn't actually play above F9. The wavetable engine doesn't go above that innately at least. So, Massive X is capable of going there in the first place.

[graspee]

lfos over nyquist

Their ships filled the skies!
They took the Earth!
Now one man has the key to fight back!

Gollanz SF 1972 1st Edition, 3rd Reprinting.

[ghettosynth]

I was wrong about the direction of the filter knob in waveshaping and what it does, sorry.

It seems to the right is "unaltered" and to the left is "filtered" and turning it to the left will even turn a square wave into a sine wave.

All this would be less confusing with a proper manual to hand.

Yeah, the missing manual is annoying. There are way too many modes with non-self-explanatory names on the oscillators. I feel like some of this could have used some editing and perhaps some of the modes could have been released later.

The (osc) filter is quite useful on a synth like this in much the same way that having filters on an FM synth is useful. Sometimes it's just easier to roll off some of the harmonics to get the sound that you're looking for. Not having to use the main filter for that, and having a separate filter per oscillator is useful. I'd like an EQ or fixed filter bank in the A/B/C section as well.

While I really like the synth, it absolutely feels like a rushed product to me. The position that the original Massive didn't display values isn't good justification for a synth released in 2019 not to display values. Whomever pointed this out is correct, NI goes out of their way on their Reaktor products to make this happen.

In some ways I think that this release reflects how much value there is to NI in the Reaktor product. I think that, despite how annoying U/I development is in Reaktor, that it really does facilitate the rapid release of products for them, same with Kontakt.

[BlitBit]

Hey guys! So I did the follow up video and I thought before I made it live to my subscribers the experts could watch it and explain to me whats happening. Ive found that the filter compounds the aliasing and the LFOs in Oscillator modes are not band-limited at all! Square waves really break this thing.

Ok, when you modulate osc with lfo (square wave)...lol...thats some nasty aliasing, sounds very bad. I hope NI will release an update soon with some oversampling options. Back to Hive, Dune 3 and Serum until MX is fixed!

Can you directly hear the output of Hive's, Dune's and Serum's LFOs and use them at audio rate? If not, what makes you think their implementation is better?

I think part of the "problem" is that Massive X uses a more modular approach than other synths, e.g. you can connect the LFO directly to the audio sink, and this now exposes some details which you cannot even see in other synths.

[HcDoom]

It doesnt matter, I for one use a lot of lfos in my sounds for my productions and I would never ever use a synth with this amount of aliasing in 2019, no matter is coming from osc or lfo modulating osc, etc etc. Its really not acceptable in this time.