PG8X (inspired by the JX8P): new beta version uploaded

[martin_l]

Hi i finalyl test this "thing" , i dont know what did you do during the making of the plug , but for the first time i must said this is the real thing , only "report" is chorus it's to "thin" comparing to mine revision of jx8p ... anyway big congrats martin !!!!!

Thanks for the kind words. I might add a 'hidden control' for the amount of modulation in the chorus.

A small status update:

I now added the automatic oversampling control, and a 'hidden' settings panel (which can be activated by a small button), which allows manual control over the oversampling, as well as a master tune. That's where I will try to put the chorus amount as well.

I aim to release this version (hopefully as the last beta) probably tomorrow night, after some more testing, and restoring the presets, which got lost during the changes.

Cheers,
Martin

[Schiffbauer]

Hi Martin,

I'm glad to hear you're finishing straight. So I am looking forward to
the last beta and eventually the final.

When the final is ready, I'm going to start my contribution patchbanks
and send it to you - 10 Banks at first - it will take a little bit time.

Unfortunately I don't have written down, what patches I used in the first 3 sounddemos - so I'll try to remember and apply all the patches from the sounddemos in the upcoming soundbanks.

ciao
armin

[martin_l]

Hi Martin,

I'm glad to hear you're finishing straight. So I am looking forward to
the last beta and eventually the final.

When the final is ready, I'm going to start my contribution patchbanks
and send it to you - 10 Banks at first - it will take a little bit time.

Unfortunately I don't have written down, what patches I used in the first 3 sounddemos - so I'll try to remember and apply all the patches from the sounddemos in the upcoming soundbanks.

ciao
armin

I am looking forward to all the patches.

I am still struggling with the compatibility problem. The changes I introduced yesterday evening, again, made the old banks unreadable. Thr problem is, that all voices are separate containers (as I am not using the built-in polyphony support of SE). So, whenever I change something, I change the first voice, then delete the others, and re-create them as copies of the new voice 1. Whenever I do this, it seems that the internal order of parameters change. That's why I want to do the release only once I know that the synth is running without any problems.

I hope that over the weekend, many people can test it on different platforms, and if it is stable, I will finally release that version, hoping that no more changes will be necessary.

Cheers,
Martin

[ermi]

I like the sound. Can't make any comparisons with hardware, though.

It makes a short noise/pop every time I load it in Renoise. (Maybe it's been mentioned, didn't read the thread.)

[martin_l]

I like the sound. Can't make any comparisons with hardware, though.

It makes a short noise/pop every time I load it in Renoise. (Maybe it's been mentioned, didn't read the thread.)

It was not yet mentioned here, but I am aware of it. However, I have not yet managed to find the source of the click. Probably, some of the filters is not initialized properly. I will have another look whether I can spot the culprit.

Cheers,
Martin

[cytone]

Is the Random LFO shape working as designed?. I can't seem to get any effect from it.

[LXNDR1]

I like the sound. Can't make any comparisons with hardware, though.

It makes a short noise/pop every time I load it in Renoise. (Maybe it's been mentioned, didn't read the thread.)

It was not yet mentioned here, but I am aware of it. However, I have not yet managed to find the source of the click. Probably, some of the filters is not initialized properly. I will have another look whether I can spot the culprit.

Cheers,
Martin

if nothing works you can set up a momentary block on the main output triggered by Monostable when the plugin loads

[martin_l]

Is the Random LFO shape working as designed?. I can't seem to get any effect from it.

Thanks for pointing this out. Indeed, I lost the connection of the noise to the S&H module during the last update. Will be fixed in the next version.

Martin

[alavux]

i wanna contribute my own patches , so i need instructions (i'm total lame for sending sysex to cubase and never needed since i write all into internal memory)

... so ...

[martin_l]

i wanna contribute my own patches , so i need instructions (i'm total lame for sending sysex to cubase and never needed since i write all into internal memory)

... so ...

Unfortunately, I have not quite figured out how to import SYSEX myself. I think it should be possible through the MIDI learn menus, but I dont know the syntax, i.e. how to markup the actual controller value. If anybody can give me a hint here, I would appreciate very much...

Furthermore, I would like you to wait for the final release version, since so far, every change I implemented made the patchbanks from a previous version unreadable. I don't know how to prevent that from happening. Hence, I would encourage you to wait until there is a version, which is confirmed to be bug free on a wide variety of hosts.


Cheers,
Martin

[IncarnateX]

Bookmarked!

Very fine synth. Great work.

[martin_l]

I like the sound. Can't make any comparisons with hardware, though.

It makes a short noise/pop every time I load it in Renoise. (Maybe it's been mentioned, didn't read the thread.)

It was not yet mentioned here, but I am aware of it. However, I have not yet managed to find the source of the click. Probably, some of the filters is not initialized properly. I will have another look whether I can spot the culprit.

Cheers,
Martin

if nothing works you can set up a momentary block on the main output triggered by Monostable when the plugin loads

I think I found the cause of the click. In order to prevent denormals in one of the filters, I shifted the signal by 1.0 (and subtracting it again after the filter), but I still initialize some internal variables of the filter to 0, which of course should be 1. That might give the click whenever the filter first starts up. I will try tonight whether that's indeed the problem.


Martin

[martin_l]

I just uploaded a new version (pg8x-62.dll), which I consider to be a release-candidate. Please, give it a try and let me know whether there are any problems. If this version works without bugs, I will make this the official release, and will also remove the restriction to 16 patches.

As I suspected in the last post, the click when loading the instrument was due to wrong initialization of the filter variables. This is fixed now.

Other new features are:

• * SETTINGS panel with control over
  • - Master Tune
    - Oscillator oversampling: MODE and QUALITY (description below)
    - Downsampler filter stages: The oversampled signal is passed through a low pass filter at roughly the Nyquist frequency. The number of stages determines the steepnes of the filter: 2 = 24dB/oct, 4 = 48 dB/oct, etc.
    - Chorus settings: adjustment of speed and depth; middle settings correspond roughly to the original.
  * The (global) oversampling is automatically adjusted according to the
  sample rate of the host. For 44.1 kHz and 48 kHz the global oversampling is 4 fold, for 88.2 kHz and 96 kHz it is reduced to 2 fold.

About the meaning of the QUALITY setting:

The waveforms are not generated from wavetables, but from a ramp generator, which is reset to 0 whenever the signal exceeds 1. The resetting to 0 much simplifies the algorithm (especially when using hard sync), but gives rise to slightly wrong frequencies. The QUALITY setting determines the internal clock of the ramp generator, and the value is multiplied on the global oversampling. E.g. for 44kHz sample rate and quality '4' the ramp generator runs at 16 fold oversampling. The 'errors' in the frequencies are more noticable at high frequencies, and in particular when ring modulating two high frequencies (giving rise to the differece of the frequencies). On the other hand, this algorithm also leads to quite low aliasing.

For 'ordinary' sounds, a manual setting of '2' or '3' is in general sufficient. For ring-modulated high pitch signals (see patch MUSICBOX), it is not. The Auto setting, adjusts the QUALITY in such way that high pitch notes (based on the larger value of the two oscillators) have a higher setting, than low picth notes. In general, this leads to higher CPU than low manual settings, but lower than a manual setting of '16'.


USER POLL:

Before releasing the final version, I would like to get your feedback on some points:

• * If you agree that 2 filter stages in the down sampler are sufficient, I can remove the user control. I personally dont hear any improvement when using a steeper filter. If you think it should stay there, I will leave it.
  
  * Would you like to keep the chorus control?
  
  * Would you like to have the key-assignment settings stored with the patch, or always default to poly (as in the hardware)?
  
  * Would you like the SETTING panel switch to be stored with the patch, or always default to the standard view?

I will wait over the weekend for replies and comments, and hope to be able to release the final version some time text week. If everybod is happy with the current state, this will be it.


Cheers,

Martin

[AdmiralQuality]

* If you agree that 2 filter stages in the down sampler are sufficient, I can remove the user control. I personally dont hear any improvement when using a steeper filter. If you think it should stay there, I will leave it.

- I can definitely hear the difference in the filter slopes. As for whether that's improvement or not is purely subjective. This is stored in the program, correct? (Not a global setting?) If so, I'd say keep it.

* Would you like to keep the chorus control?

- Why not? It's saved with each program, correct? Control is good.

* Would you like to have the key-assignment settings stored with the patch, or always default to poly (as in the hardware)?

- With the patch. Programs are usually designed with a certain performance style in mind, so best not to model that deficiency of the original. Also, it's vital that the host recall its project the same way. So if it's not stored in the patch/program, does that mean it will reset to default when a new instance is created on loading a previously saved host project? That would be bad.

* Would you like the SETTING panel switch to be stored with the patch, or always default to the standard view?

- Default to standard. (Although the settings switch should change its label to "performance" or something when depressed, to reflect that its now hiding the detune and assignment mode controls from you.)

Great work Martin!

[Schiffbauer]

I've tested your pre-Final and it's great. Made some sounds to attempt -
so that it is.

About your questions - "Admiral" said in my view.
It's all said and done. As far as I concerned, you can make this
version as official release - you'd only remove the restriction
to 16 patches.

A really perfect and good work, Martin......

ciao
armin