SynthTone PWM - aliasing?

Forum for JUPITER-80
knolan
Posts: 158
Joined: 19:50, 10 April 2008

SynthTone PWM - aliasing?

Post by knolan »

Has anyone else noticed that if you use PWM in a SynthTone, it introduces horrific aliasing in the higher octaves?

To me, it's brutal, to the point of regarding it as a fatal flaw and unusable.

To demonstrate, load Synth Legends, go into single play mode, choose Liveset 423 - SLO-Jn106 B Str and play D5 (on the keyboard) and hold it. Can you hear the aliasing almost as loud as the string sound itself!

Anyone got a work around / solution?


Kevin.
Dany2
Posts: 6
Joined: 01:40, 4 March 2014

Re: SynthTone PWM - aliasing?

Post by Dany2 »

knolan wrote:Has anyone else noticed that if you use PWM in a SynthTone, it introduces horrific aliasing in the higher octaves?

To me, it's brutal, to the point of regarding it as a fatal flaw and unusable.

To demonstrate, load Synth Legends, go into single play mode, choose Liveset 423 - SLO-Jn106 B Str and play D5 (on the keyboard) and hold it. Can you hear the aliasing almost as loud as the string sound itself!

Anyone got a work around / solution?


Kevin.
You don't need to load a specific synth tone in order to hear this flaw of the two PWM wave variations. Just take one single synth tone partial and choose the PWM wave, increase the PW amount and then choose wave variation 1 or 2...then play in higher ranges.

Partial 1 and 2 of the mentioned synth tone are both using the second variation of the PWM wave, indicated by the green virtual wave select button (first variation is red).

If you change those synth partials to the basic PWM wave, the disturbing effect will almost disappear, but still remain slightly audible in higher ranges.

One really wonders why the sound programmer has decided to use the flawed variation instead of the basic PWM wave, as variations 1 and 2 are only usable for bass sounds, unless you want to create some strange 50s retro sci-fi sounds...

We have to ask Roland some questions:

Why do you allow the selection of two completely flawed variations of the PWM wave, if the basic PWM wave has already a dangerous tendency for aliasing in higher ranges?
Are your engineers deaf?
Are your sound programmers deaf?
When will you release an OS Version3 update for your alleged flagship synth, which fixes this obvious and unacceptable engineering failure?


This absolutely unacceptable flaw has to be present on any Roland product with the SN Synth engine, like e.g. Integra-7 or FA Workstation...Or is it just the JP-80 flagship?

The prophet turned out to be false...

Thanks to Kevin for this eye-opener!

Image

PS: I am so glad, about the low aliasing and flawless OASYS!
User avatar
Grammar Wombat
Posts: 435
Joined: 01:52, 23 July 2008
Location: Here

A cat and a PWM!

Post by Grammar Wombat »

This is a most disturbing discovery.

We already knew that Roland's flagship synthesizer lacked any form of mouse support…

Image

(Harry still has nightmares about that incident.)

Now, we learn that the Jupiter-80 does support a cat!

Cat support!?
Dany2 wrote: Image
Harry is absolutely livid at this cruel affront!

And I must say that the sound of a PWM (Particularly Wrathful Mouse) is quite unbearable.
A Harry Situation.jpg
A Harry Situation.jpg (24.82 KiB) Viewed 5305 times
User avatar
PauloF
Posts: 4201
Joined: 02:35, 16 January 2006
Location: Lisbon, Portugal
Contact:

Re: A cat and a PWM!

Post by PauloF »

Grammar Wombat wrote:This is a most disturbing discovery.

We already knew that Roland's flagship synthesizer lacked any form of mouse support…

Image

(Harry still has nightmares about that incident.)

Now, we learn that the Jupiter-80 does support a cat!

Cat support!?
Dany2 wrote: Image
Harry is absolutely livid at this cruel affront!

And I must say that the sound of a PWM (Particularly Wrathful Mouse) is quite unbearable.
A Harry Situation.jpg
LOL LOL LOL
User avatar
PauloF
Posts: 4201
Joined: 02:35, 16 January 2006
Location: Lisbon, Portugal
Contact:

Re: SynthTone PWM - aliasing?

Post by PauloF »

Dany2 wrote:
knolan wrote:Has anyone else noticed that if you use PWM in a SynthTone, it introduces horrific aliasing in the higher octaves?

To me, it's brutal, to the point of regarding it as a fatal flaw and unusable.

To demonstrate, load Synth Legends, go into single play mode, choose Liveset 423 - SLO-Jn106 B Str and play D5 (on the keyboard) and hold it. Can you hear the aliasing almost as loud as the string sound itself!

Anyone got a work around / solution?


Kevin.
You don't need to load a specific synth tone in order to hear this flaw of the two PWM wave variations. Just take one single synth tone partial and choose the PWM wave, increase the PW amount and then choose wave variation 1 or 2...then play in higher ranges.

Partial 1 and 2 of the mentioned synth tone are both using the second variation of the PWM wave, indicated by the green virtual wave select button (first variation is red).

If you change those synth partials to the basic PWM wave, the disturbing effect will almost disappear, but still remain slightly audible in higher ranges.

One really wonders why the sound programmer has decided to use the flawed variation instead of the basic PWM wave, as variations 1 and 2 are only usable for bass sounds, unless you want to create some strange 50s retro sci-fi sounds...

We have to ask Roland some questions:

Why do you allow the selection of two completely flawed variations of the PWM wave, if the basic PWM wave has already a dangerous tendency for aliasing in higher ranges?
Are your engineers deaf?
Are your sound programmers deaf?
When will you release an OS Version3 update for your alleged flagship synth, which fixes this obvious and unacceptable engineering failure?


This absolutely unacceptable flaw has to be present on any Roland product with the SN Synth engine, like e.g. Integra-7 or FA Workstation...Or is it just the JP-80 flagship?

The prophet turned out to be false...

Thanks to Kevin for this eye-opener!

Image

PS: I am so glad, about the low aliasing and flawless OASYS!
No, unfortunately is not only the Jupiter-80/50... The integra has exactly the same problem. Don't understand as why a 2011 synth has that so evident issue.
I know that some waveforms tend to create undesirable sub-harmonics (like the square wave) but It's not difficult to design no-aliasing digital oscillators and there are several techniques to cut those frequencies out...
knolan
Posts: 158
Joined: 19:50, 10 April 2008

Re: SynthTone PWM - aliasing?

Post by knolan »

Thanks Dany for your post - very useful. It beggars belief that the synth Legend Programmers saw fit to include this in their Juno 106 string emulations - couldn't they hear it?

I own a Juno106 for a long time and know that instrument inside out - and I have to say - aliasing aside - some of those Juno106 live sets are quite good indeed!there's a particular 'dry-ness' to the juno106 sound when filter resonance is fully off, and some of these sound have that quality -it's quite impressively similar!


Thanks,
Kevin
skinmechanic
Posts: 205
Joined: 13:39, 14 June 2004
Location: UK, Leeds
Contact:

Re: SynthTone PWM - aliasing?

Post by skinmechanic »

I just don't think they have the money or R&D time to spend on the JP range or it's variants. The Synth Ledgends has also been abandoned from it's initial plan of releasing new volumes every six months.

I find the later roland models in the Synth Ledgends specifically the D50 tones were pretty far off the mark. Specially when how close Korg got with their M1 and Triton sounds for the Krome.

I think the cost of R&D for the Jupiter range from the beginning far outweighs their return, which is why they are re-marketing it as a mid range workstation in the FA's
Dany2
Posts: 6
Joined: 01:40, 4 March 2014

Re: SynthTone PWM - aliasing?

Post by Dany2 »

knolan wrote:Thanks Dany for your post - very useful. It beggars belief that the synth Legend Programmers saw fit to include this in their Juno 106 string emulations - couldn't they hear it?

I own a Juno106 for a long time and know that instrument inside out - and I have to say - aliasing aside - some of those Juno106 live sets are quite good indeed!there's a particular 'dry-ness' to the juno106 sound when filter resonance is fully off, and some of these sound have that quality -it's quite impressively similar!


Thanks,
Kevin
I am finally wondering about myself, that I wasn't aware of the PWM wave's variation aliasing issue. But so far, I've only used the basic PWM wave and never its two variations for my own sound programs.

After this quite "shocking revelation", I've tested all the other vintage modeled analog waveforms, and especially their two variations, concerning any aliasing issues. Fortunately there aren't any flaws audible!

Only the PWM wave seems to be frail in this regard. The aliasing of the basic PWM wave is acceptable and only slightly audible in high ranges. However the basic PWM wave version still reveals sligthly more aliasing, than all the other vintage modeled analog waveforms.

The other two PWM wave variations have such a huge amount of aliasing artifacts in the higher ranges, that one can use those artifacts on purpose as ingredients for strange digitally crushed sounds. This was possibly the reason, why the engineers have left the option to choose those crushed PWM wave variations anyway, but it should have been mentioned in the user manual.

The PWM wave is also the only waveform, which doesn't react to the Wave Shape function.

The Jupiter-80 is such a rewarding musical instrument and such a sheer joy to program and to play, that playing it just calmed me down from my initial "PWM aliasing shock" and the resulting rant in my previous post.

"The quarrel of lovers is the renewal of love"

Image
knolan
Posts: 158
Joined: 19:50, 10 April 2008

Re: SynthTone PWM - aliasing?

Post by knolan »

Excellent point Dany.

My first ever keyboard was the Casio 1000P - it was an advance by Casio in the early 80's in that it offered 1000 sounds, and an intriguing sequencer (which I still use today because it has a unique feature that has never appeared on any other sequencer making it a brilliant 'live performance' sequencer).

but that was in the earliest days of digital, and the Casio sounds all suffered from horrendous aliasing in the upper octaves - which I used (and still use) to produce a plethora of chime and bell sounds, greatly extending the casio's palette of sounds - especially for rhythmic and bell sounding sequences.
User avatar
Saxifraga
Posts: 173
Joined: 17:23, 8 January 2014
Location: Berlin

Re: SynthTone PWM - aliasing?

Post by Saxifraga »

There is some basic understanding of signal processing theory missing here. It's not the flaw of any engineer but a feature/bug of nature itself!
Any square wave or PWM wave needs an infinite amount of harmonic and inharmonic partials.
According to nyqist theorem you cannot reproduce truthfully any signal that has partials beyond higher than half the sampling frequency. That's normaly 44.1 kHz. So all partials beyond 22.5 kHz get maped to artificial aliasing frequencies. The higher the note the shorter the square the higher the amplitudes of high frequency partials.

PWM is a nightmare in a finite world.

Analog synth have no problem with that because they don't have real mathematicaly exact PWM source waves!
PWM generated in analog gear has from the biginning not enough partials to create aliasing because the square is constructed from a finite number of partials!

So the digital PWM wave is much nearer to the exact mathematical thing. It's a feature not a bug of th JP-80 to give you those waveforms for experiments. Just apply a filter to those waves that cut the aliasing away and you have an analog PWM wave. Voila!
knolan
Posts: 158
Joined: 19:50, 10 April 2008

Re: SynthTone PWM - aliasing?

Post by knolan »

But, Korg, for example, provide alias free PWM on their PolySixEx synth as implemented on the OASYS as far back as 2007. Surely this is achievable by a Roland today - just run the sampling rate high enough and filter the aliasing harmonics.


It's poor design.
knolan
Posts: 158
Joined: 19:50, 10 April 2008

Re: SynthTone PWM - aliasing?

Post by knolan »

Unfortunately this issue is prevalent across all of the PWM options available.

I've been going though the Synth Legends Livesets to see which ones I'm interested in , and a significant number of them suffer from Aliasing in the upper part of the keyboard.


I've submitted an email to tech support at Roland UK to argue that this needs a fix. I'll let you know what the response is. I might urge as many of you as possible to send a similar email.


This is a serious flaw in the synth tone engine that will similarly impede the JP50, Integra-7, FA range and perhaps the System 01 too (or at least its reputation of this flaw becomes more known).

Overall, there is so much that's good about this synth engine, but this issue is surely problematic across a large range of Roland's current synthesisers.


Kevin.
knolan
Posts: 158
Joined: 19:50, 10 April 2008

Re: SynthTone PWM - aliasing?

Post by knolan »

Apologies for a third straight post on this - but -

This is shocking. just about every synth tone Liveset in the entire Synth Legacy collection has brutal aliasing in the upper octaves.

Virtually every one of them is useless - totally useless for recording purposes.


What were Roland thinking? I honestly cannot believe how bad this is. I use the JD800, and XV5080 - even the JV2080 - none have this issue.

For me, this if fatal regarding using the Synth Tone engine. I just cannot believe it. It's so bad it beggars belief. Just how in the hell did this get past quality control in Roland? It seems to me that this synth engine was designed by rookies who don't know any better. Didn't they hear it?? And - who are the programmers who designed the Synth Legends sound set? Weren't they aware of it? Why did they use PWM - who's the target user?

If the same synth engine in the FA range, Integra and so on - it's horrendous.

I'm flabbergasted. And I'm so disappointed. For all my griping about the ergonomics and OS, I had regarded the JP80 as a pinnacle of sound quality. This is true for the Acoustic Tones. But this issue renders the Synth Tone engine essentially useless in any professional context. It just can't be used, because virtually every live set uses some synth tone with PWM. Hugely, hugely disappointing.
knolan
Posts: 158
Joined: 19:50, 10 April 2008

Re: SynthTone PWM - aliasing?

Post by knolan »

Also,

The Supersaw wave is not a VA wave - it's a sample. If you isolate it from all effects in just one tone and play it in the upper octaves, you can hear the looping (and - Aliasing!).

Furthermore, the Square wave also suffers from Aliasing - Try Synth Legends Liveset 179 and activate only Synth Tone number 1 - 0029 SLT-JP6 6 - it exhibits a lot of alining and does not use PWM

This goes from bad to worse!
User avatar
PauloF
Posts: 4201
Joined: 02:35, 16 January 2006
Location: Lisbon, Portugal
Contact:

Re: SynthTone PWM - aliasing?

Post by PauloF »

knolan wrote:Also,

The Supersaw wave is not a VA wave - it's a sample. If you isolate it from all effects in just one tone and play it in the upper octaves, you can hear the looping (and - Aliasing!).

Furthermore, the Square wave also suffers from Aliasing - Try Synth Legends Liveset 179 and activate only Synth Tone number 1 - 0029 SLT-JP6 6 - it exhibits a lot of alining and does not use PWM

This goes from bad to worse!
There is some aliasing in some waves yes, specially the ones that generate more harmonics, but to my hears the "horrendous" effect only happens for C7 (8.372 hz) and above...not on "normal" frequencies

Thank God that I'm deaf ;-)
Post Reply