Repro-1 (out now)

[Urs]

I'll be off for a few days, but please keep it coming!

@Vesius You're right, what you linked in the Soundcloud demo is most definitely aliasing. I think it results from differences in volume, such that the waveshaper is hit harder in some "louder" models. The waveshaper isn't a cosy tanh() shape but a linearised one, which looks a bit like this plot. This can be compensated for later on, but it would feel like cheating to me at this stage.

Another note on volume: If I'm not completely mistaken, the analogue filter output can not exceed a certain volume relative to max VCO gain, despite each stage being unbounded. I have set this point to +/- 1.0 aka 0dB. None of the filter models including the most CPU expensive one keeps perfectly within that range. Some may even blow up to infinity. Hence, in order to keep some models stable at all I have added a safe guard at +24dB or +/-16.0. When this gain is crossed, all filter states are flushed. This may be someone's "harsh" or "aliasing" but someboday elses "alive" and "interesting". It is however not an analogue artifact (even though it might be similar to a power breakup when more current is drawn from an OpAmp than is available, but this doesn't happen in the original circuit)

Once I'm back I might do a video with some more noodling on the Pro-One so people get a feel for its sound.

- U

[moscom_electronics]

Test Repro.zip

A quick go on an analogue Pro-One, C3 then C5

http://www.u-he.com/urs/ProOneFilterFMSweep.mp3

ProOneFilterFMSweep.wav

Those second of those two audio samples just reinforces my belief that it's 1 or 4. That is almost exactly the sort of sound I was hearing when sweeping the filter on the settings I had whilst testing. Filters 2,3 and 5 didn't produce that sort of sound. 3 started to but it disappeared very quickly as I opened the filter.

The last little flourish on the second sample sounds like something from Pacman.

I f I may, Urs did this test following this post of mine:

http://www.kvraudio.com/forum/viewtopic ... 0#p6431179

where I used a patch, that you can download, to demonstrate RePro Alpha. This post also contains the audio examples I have uploaded to Soundcloud. The results are NOT affected by degradations possibly induced by Soundcloud's compression algorithms. They sound the same on my computer.

To sum up again, these audio examples show that 1 produces a lot of digital-like sound (that can be cool in some context) and 4 produces artifacts (that do not sound like aliasing, just some sort of stepping and very strange spatial effects, that I find actually interesting ): both are respectively very and slightly different from the actual Pro One. 3 is the smoothest one and the closest to Urs's test on the actual Pro One. 2 and 5 just do not produce the same sound as both the actual Pro One and the other software filters. Note that there is no personal interpretation here, this is just factual listening.

I am not sure that this makes 3 the most analogue of all the filters. It just makes it the best behaved even under extreme settings, the one that still responds under extreme settings (2 and 5 just collapse, and produce no interesting harmonic content, making them dull and boring) without produce a significant amount of artifacts (in contrast to 4 and 1) and so far as I can tell, the best one to my ears, as I can expect it to remain musical whatever the modulation paths I set up. Also, 3 is my filter of choice to dial in all those sounds that I have heard on records (Drexciya, Yazoo, etc...) that I knew were coming from a ProOne, and from this point of view, this is definitely the most analogue-sounding one.

Sorry for this third post regarding my demo, but I wanted things to be correct regarding Urs's demo.

[moscom_electronics]

I added some PWM via max to make a patch to see what this will sound like once it heads to full beta/release.

This does have a lot of the character of the Pro One with this sort of sound




I also went through some paces of the standard self-oscillation with keyboard tracking I like to do with synths to get more familiar with their filter, portamento, etc. Not all synths apply portamento to their filter keyboard tracking. Bravo Urs!

Really cool!

[bcm70]

@moscom - your test is interesting. What settings did you use for it?

In your example filter 3 holds on a lot better than it did in my testing. It died off a lot more like 2 and 5 have in your soundcloud examples, which is why I lent towards 4.

Looks like I'll have to play around a bit more.

[hakey]

Which model sounds closest to this?

[Urs]

Which model sounds closest to this?

Which btw. none of the recordings people sent sound like - most of them had filter frequency go down with higher resonance

[Vesius]

thanks for the clarification Urs.

tanh() in the feedback loop is a cpu mangler, so if this is where you are aiming for a less CPU intense filter, I think a lot of people are going to be extremely happy given how popular filter1 is.

[itoa]

Nice stuff!
I think that 1st filter is the closest. This begs for a bad PSU model

btw. if you like analog filters lets try Dagger by BeepStreet as well (4 circuit modelled, 8x oversampled filters with tons of nonlinearities)

[moscom_electronics]

@moscom - your test is interesting. What settings did you use for it?

In your example filter 3 holds on a lot better than it did in my testing. It died off a lot more like 2 and 5 have in your soundcloud examples, which is why I lent towards 4.

Looks like I'll have to play around a bit more.

You can download the patch if you follow the link in my message. It is a zip file that is at the top of the message. Thanks!

[hakey]

Which model sounds closest to this?

Which btw. none of the recordings people sent sound like - most of them had filter frequency go down with higher resonance

We'll, I guess 3 is the 'accurate' model, then.

It's interesting that the different models need pushing right to the extremes for the differences to become obvious. To my ears, most typically useful synth patches sound very close from one model to the next.

[beely]

It's interesting that the different models need pushing right to the extremes for the differences to become obvious. To my ears, most typically useful synth patches sound very close from one model to the next.

In general yes, it's the extremes where the cheaper models start to fall down...

[VirusTI82]

I want the most analogue sounding filter. (I took Filter 4)
It doesn't matter for me how much CPU is needed.
I just want the best analogue sounding RePro One as possible!
People who got a weak CPU should buy a better one or don't use RePro One.

People who got not minimum a Intel i7 CPU should not make music with a computer!

[Urs]

Don't worry about CPU too much. It'll be fine no matter what algorithm we choose.

[feriado]

For me, 3 is the best.
I feel digital taste with 2 and 5 especially when the resonance is middle or high and the VCF is modulated by lower frequency and deep.
I feel 1 and 4 are good, but little thin tones.

[Robmobius]

Think I'll go with 4, just sounds a wee bit nicer than the other ones.

God, I can't wait for this to be released. Those raw Oscillators are lovely and stacked they sound immense.

I need it now! This very minute...