SynthTone PWM - aliasing?

Forum for JUPITER-80
User avatar
kimsnarf
Posts: 275
Joined: 17:55, 4 January 2013
Location: Oslo, Norway

Re: SynthTone PWM - aliasing?

Post by kimsnarf »

knolan wrote:I have to also say that I think its a sign of the absolute shite quality of current pop/dance recordings if JP80 owners are using these patches and not even noticing the effect. To paraphrase John Cleese - it's BLEEDING OBVIOUS to me. Clearly the vast majority of very high profile JP80 users out there have horrendously bad ears and judgement.
If you go looking for imperfections you will find them. I think most people learn to accept flaws, even in expensive products. If the majority of the sounds on the Jupiter-80 were obviously useless due to noise, this would be a critical issue. But apparently most users of this synth have yet to notice the aliasing and are happy playing the instrument. The aliasing can't be that prevalent in use, at least not for musical purposes.

If you want to convince others that this is a major issue in practice, I suggest making a few musical examples where the aliasing is obvious. If 80% of the sounds are affected that should be easy.
arfy
Posts: 19
Joined: 21:38, 27 February 2013

Re: SynthTone PWM - aliasing?

Post by arfy »

Hi all.
Well, I'm also quite a VA fan, especially PWM waves, so I want to hear now just how bad this issue is. Would someone be able to provide samples of what the jp80/50/I7/FA's doing with pwm's, and then something that's considered better? Have the sample in lossless form if possible, so that lossy artifacts aren't introduced as well? Yes, I know what aliasing sounds like, and it can be used creatively if you know how. so, come on, step up and let's actually hear what's going on, for those of us who don't own a roland SNS synth. :)
JunoJohn
Posts: 815
Joined: 08:29, 24 February 2008
Location: Washington DC

Re: SynthTone PWM - aliasing?

Post by JunoJohn »

arfy wrote:Would someone be able to provide samples of what the jp80/50/I7/FA's doing with pwm's, and then something that's considered better?
Here are some samples from my Integra-7 along with an MKS-50, a DCO analog synth.

The waveforms are played in the following order in each sample:

1) MKS-50: PULSE 03
2) Integra-7 SN-S: PW-SQR A
3) Integra-7 SN-S: PW-SQR B
4) Integra-7 SN-S: PW-SQR C

Each sample is played at a different pitch. C6 is two octaves above middle C.

You will hear that PW-SQR B and C are the waveforms with the serious aliasing problem.
C6.mp3
(487.78 KiB) Downloaded 876 times
C7.mp3
(487.78 KiB) Downloaded 635 times
C8.mp3
(487.78 KiB) Downloaded 681 times
If anyone wants the AIFF files, let me know and I will upload them.
rrricky rrrecordo
Posts: 2
Joined: 04:36, 19 August 2014

Re: SynthTone PWM - aliasing?

Post by rrricky rrrecordo »

Hi all,

Picked up a VR-09 which is a sweet little board. The synth section is identical to that in JP-80 V1 albeit only two layers (six partials). I've put the virtual waveforms up on an oscilloscope (with variations) and there are some disturbing findings for me.

Most disturbing: white and pink noise are samples on VR-09. Loops are clearly discerned, particularly when highs are cut. Is this a cost cutting "feature" on the 09? I ask because the iPad editors for VR-09's and JP-80 are virtually identical apart from JP-80's four LPFs.

Curious only because I am ready to pick up a used 80 at a nice price but I do expect analog models of the essential waveforms. Oddly, some sampled waves in the PCM section behave quite unlike samples, some sync waves in particular.

So, main waveforms analog modeled or sampled? Listen to noise gens carefully

Thanks

rr
User avatar
Tschury
Posts: 9
Joined: 23:56, 8 August 2014

Re: SynthTone PWM - aliasing?

Post by Tschury »

rrricky rrrecordo wrote:Hi all,

Picked up a VR-09 which is a sweet little board. The synth section is identical to that in JP-80 V1 albeit only two layers (six partials). I've put the virtual waveforms up on an oscilloscope (with variations) and there are some disturbing findings for me.

Most disturbing: white and pink noise are samples on VR-09. Loops are clearly discerned, particularly when highs are cut. Is this a cost cutting "feature" on the 09? I ask because the iPad editors for VR-09's and JP-80 are virtually identical apart from JP-80's four LPFs.

Curious only because I am ready to pick up a used 80 at a nice price but I do expect analog models of the essential waveforms. Oddly, some sampled waves in the PCM section behave quite unlike samples, some sync waves in particular.

So, main waveforms analog modeled or sampled? Listen to noise gens carefully

Thanks

rr
The JP-80 and VR-09 synth sections are certainly not identical, just similar looking. The similar look of the editors also doesn't mean much.

The JP-80's VA Synth engine offers classic analog modeled Oscillator waveforms: http://www.roland.com/synth/JUPITER/

But there is also an option to choose from 363 PCM waveforms as Osc source for the VA synth engine.

Image

As you can see, the JP-80 offers a dedicated noise generator for pink and white noise (Variation button). But there are also two PCM waveforms available for pink and white noise. (PCM waveforms No.362 and 363)

I've just compared the noise versions and it is evident that the JP-80's noise generator is indeed a genuine noise generator and not just a looped sample, while the JP-80's PCM noise waveform versions are obviously just sampled noise with an audible loop, which are probably exactly the same looped noise samples which you can hear on your VR-09!

So yes, the Jupiter-80 has a genuine noise generator and the VR-09 obviously doesn't... :)

The JP-80 has no Osc sync. So the sync waveforms are indeed just PCM waveforms. But they are very good and usable.

Instead, the JP-80 offers a Wave Shape function, which is a kind of Cross Modulation and it also offers Ring Modulation.

The following is an audio example of the JP-80's Wave Shape function. First you hear an arpeggiated single JP-80 Osc with a basic VA saw wave (Filter bypass, no effects) with increasing Wave Shape value. Second you hear the same thing with two Oscs (2 Partials), which results in a different effect. Third you hear the same thing for one Osc, but with a PCM waveform as Osc source instead of the VA Saw wave:
JP-80 Wave Shape function - mp3.mp3
(1.39 MiB) Downloaded 280 times
And don't forget: the JP-80 offers up to 30 VA synth Partials per note, compared to only 6 Partials on the VR-09...

Good luck with your decision!
rrricky rrrecordo
Posts: 2
Joined: 04:36, 19 August 2014

Re: SynthTone PWM - aliasing?

Post by rrricky rrrecordo »

Hey thanks Tschury!!!

The two are similar enough that MusicRadar says the synth engines in JP-80 and VR-09 are the same.

http://www.musicradar.com/reviews/tech/ ... 09-581965/

Even the Pulse Wave alias problem exists on VR-09 in the top two octaves,very strange as it's very much like an LFO running at a consistent frequency across this two octave range... except for the D and E two octaves above Middle C, which differ in frequency (and waveshape) from the aliasing of all the other notes within this range.

Btw VR-09 sports the same waveshaping function and all 363 PCM waves of the JP-80 (I have had all 363 up on the scope and all are very well done indeed). As previously mentioned, virtually every single synth parameter in the VR-09 app is similar to those appearing on the JP-80 app (indeed, minus the JP-80's four additional LPF options the two synth apps are not similar but in fact the same - apart from the number of possible layers and respective orange and blue themes).

Again with the Noise generator, the original "wave" appears to be random but I have had discerning students pick out the loop. The VR-09's noise loops (in the Waveform Selector as opposed to PCM 362 and 363) are easily heard by closing the filter a bit more than half way - say 11 o'clock - and maxing up the resonance. When the alternate Noise waves are selected, the second choice (red) is darker and one can clearly detect a "railroad crossing" type loop. The third Noise option (green) on VR-09 just seems to be digital spikes, nothing I can use... sure hope the JP is a different story!

In any case, I will take some decent headphones when I meet with the vendor. Thanks for your input and the prompt response :)

rr
User avatar
Tschury
Posts: 9
Joined: 23:56, 8 August 2014

Re: SynthTone PWM - aliasing?

Post by Tschury »

rrricky rrrecordo wrote:Hey thanks Tschury!!!

The two are similar enough that MusicRadar says the synth engines in JP-80 and VR-09 are the same.

http://www.musicradar.com/reviews/tech/ ... 09-581965/

Even the Pulse Wave alias problem exists on VR-09 in the top two octaves,very strange as it's very much like an LFO running at a consistent frequency across this two octave range... except for the D and E two octaves above Middle C, which differ in frequency (and waveshape) from the aliasing of all the other notes within this range.

Btw VR-09 sports the same waveshaping function and all 363 PCM waves of the JP-80 (I have had all 363 up on the scope and all are very well done indeed). As previously mentioned, virtually every single synth parameter in the VR-09 app is similar to those appearing on the JP-80 app (indeed, minus the JP-80's four additional LPF options the two synth apps are not similar but in fact the same - apart from the number of possible layers and respective orange and blue themes).

Again with the Noise generator, the original "wave" appears to be random but I have had discerning students pick out the loop. The VR-09's noise loops (in the Waveform Selector as opposed to PCM 362 and 363) are easily heard by closing the filter a bit more than half way - say 11 o'clock - and maxing up the resonance. When the alternate Noise waves are selected, the second choice (red) is darker and one can clearly detect a "railroad crossing" type loop. The third Noise option (green) on VR-09 just seems to be digital spikes, nothing I can use... sure hope the JP is a different story!

In any case, I will take some decent headphones when I meet with the vendor. Thanks for your input and the prompt response :)

rr
I was just optimistically guessing in my previous, somewhat redundant post. But as we've found out in this thread, the SuperSaw appears to be sample based. So it wouldn't surprise me, if the JP-80 noise generator is a "fake" as well, although my ears cannot detect an audible noise loop, like in the case of the PCM noise wavs.
Post Reply