Repro-1 (out now)

[electro]

One thing I've noticed, and that RePro has brought new members and old members with very few posts back to KVR! It sure has some magic

Yeah, noticed that too!

That should give you a good market indicator of the potential popularity of a parameter accurate 5 voice U-he Prophet 5 VSTI. I was expecting to see it get done by NI's Monark team for Komplete 10 but it wasn't to be.

I don't understand why you ask the public for filter opinions instead of just doing empirical tests between several Pro Ones. This is the way Cytomic tweaked his filters for Drop.

[Urs]

One thing I've noticed, and that RePro has brought new members and old members with very few posts back to KVR! It sure has some magic

Yeah, noticed that too!

That should give you a good market indicator of the potential popularity of a parameter accurate 5 voice U-he Prophet 5 VSTI. I was expecting to see it get done by NI's Monark team for Komplete 10 but it wasn't to be.

I don't understand why you ask the public for filter opinions instead of just doing empirical tests between several Pro Ones. This is the way Cytomic tweaked his filters for Drop.

Well, for years you've been asking me to do a simple monosynth. Did I read this right that now you're asking for a polysynth?

The research done here is not about differences between several Pro-Ones. That is not what this is about!

Andy and I are friends. We help each other out. I don't know what you want to say or insinuate by mentioning him since this thread isn't about modeling methodology, it is about mathematical vs. perceptual differences within a single model. It says nothing about the quality, diversity and accuracy of any final product.

[EnGee]

One thing I've noticed, and that RePro has brought new members and old members with very few posts back to KVR! It sure has some magic

Yeah, noticed that too!

IMO, it indicates the success of the synth and the way it is presented.

To be honest, I loaded Monark and compared RePro with it since they are in the same category and built my vote based on it (because Monark is the most analog I have).

Of course they are very different but the level of the sound quality is the same. I just hope there will be advanced options like the B view of Monark

[electro]

One thing I've noticed, and that RePro has brought new members and old members with very few posts back to KVR! It sure has some magic

Yeah, noticed that too!

IMO, it indicates the success of the synth and the way it is presented.

To be honest, I loaded Monark and compared RePro with it since they are in the same category and built my vote based on it (because Monark is the most analog I have).

Of course they are very different but the level of the sound quality is the same. I just hope there will be advanced options like the B view of Monark

I do not believe in comparing two synths based on different hardware schematics. It only makes sense to me to be comparing something like Monark to other Minimoog emulators, preferably at extreme settings where the shortcomings of virtual models become more apparent.

[electro]

One thing I've noticed, and that RePro has brought new members and old members with very few posts back to KVR! It sure has some magic

Yeah, noticed that too!

That should give you a good market indicator of the potential popularity of a parameter accurate 5 voice U-he Prophet 5 VSTI. I was expecting to see it get done by NI's Monark team for Komplete 10 but it wasn't to be.

I don't understand why you ask the public for filter opinions instead of just doing empirical tests between several Pro Ones. This is the way Cytomic tweaked his filters for Drop.

Well, for years you've been asking me to do a simple monosynth. Did I read this right that now you're asking for a polysynth?

The research done here is not about differences between several Pro-Ones. That is not what this is about!

Andy and I are friends. We help each other out. I don't know what you want to say or insinuate by mentioning him since this thread isn't about modeling methodology, it is about mathematical vs. perceptual differences within a single model. It says nothing about the quality, diversity and accuracy of any final product.

my last post seems to have gotten lost. I was just saying I don't really think you need any outside opinions when it comes to emulating filter design, just blind listening tests with a high fail rate.

[beely]

I was just saying I don't really think you need any outside opinions when it comes to emulating filter design, just blind listening tests with a high fail rate.

He doesn't need outside opinions to implement filters. He wants to get an idea of what regular people's consensus is on the level of accuracy needed. He *knows* which are *technically* the better filters, he wants to see if we really require the highest level of accuracy, or if some lower level is just fine.

[wagtunes]

I was just saying I don't really think you need any outside opinions when it comes to emulating filter design, just blind listening tests with a high fail rate.

He doesn't need outside opinions to implement filters. He wants to get an idea of what regular people's consensus is on the level of accuracy needed. He *knows* which are *technically* the better filters, he wants to see if we really require the highest level of accuracy, or if some lower level is just fine.

Exactly, which is why he didn't just come out and tell us which filter is which right from the start.

[Urs]

I was just saying I don't really think you need any outside opinions when it comes to emulating filter design, just blind listening tests with a high fail rate.

He doesn't need outside opinions to implement filters. He wants to get an idea of what regular people's consensus is on the level of accuracy needed. He *knows* which are *technically* the better filters, he wants to see if we really require the highest level of accuracy, or if some lower level is just fine.

It's that, but even more than that.

Since Diva's release we suffered stress from two directions: There was some from the side of the people who insisted it didn't sound analogue. Those were easily ignored. Then there were the people who thought it took too much CPU, and that wasn't warranted. This thread is about latter. It's about showing that implementing the same physical/analogue model sounds different if done with CPU saving shortcuts.

What we're testing is something akin Diva's Divine mode versus 4 gradually more CPU saving methods. According to those stressy people, these methods are supposed to be indistinguishable from, say, 8x oversampling. And yet, they are not.

So, whenever someone may come and say "your stuff isn't optimized", I'll just link this thread

[dbone]

I was just saying I don't really think you need any outside opinions when it comes to emulating filter design, just blind listening tests with a high fail rate.

He doesn't need outside opinions to implement filters. He wants to get an idea of what regular people's consensus is on the level of accuracy needed. He *knows* which are *technically* the better filters, he wants to see if we really require the highest level of accuracy, or if some lower level is just fine.

It's that, but even more than that.

Since Diva's release we suffered stress from two directions: There was some from the side of the people who insisted it didn't sound analogue. Those were easily ignored. Then there were the people who thought it took too much CPU, and that wasn't warranted. This thread is about latter. It's about showing that implementing the same physical/analogue model sounds different if done with CPU saving shortcuts.

What we're testing is something akin Diva's Divine mode versus 4 gradually more CPU saving methods. According to those stressy people, these methods are supposed to be indistinguishable from, say, 8x oversampling. And yet, they are not.

So, whenever someone may come and say "your stuff isn't optimized", I'll just link this thread

Thanks a million, Urs, for having one of the most thoughtful and transparent development methodologies in the audio world - or any other. Of course, it's the end result that is the most important but tests like this really help to advance the discussion around audio software. I must confess to being one of those tedious purists - mainly because I was around when most of these original instruments came out and I'd love to own the originals. Of course, I realize that new models (like Diva) can be even more inspiring, but for me, it's just such a shame when more time is spent getting the UI to look like the original and so much of the original character is missing. It's so cool to expose the inner workings of what's going on with the development. I think that left to their own devices, u-He would continue to lead the way in amazingly beautiful sounding instruments. It's even better that they bring users along for the ride from conception to completion. I don't mean to sound like a kiss-ass, but the way they do things is really impressive to me.

[Funkybot's Evil Twin]

Since Diva's release...there were the people who thought it took too much CPU, and that wasn't warranted. This thread is about latter. It's about showing that implementing the same physical/analogue model sounds different if done with CPU saving shortcuts.

Has there been a lot of "CPU too high" complaints with Diva lately? I know it was an issue on first release, but once you implemented the multi-core button, it's been 100% a non-issue here. On my newer i7 desktop, even less so. I consider Diva's CPU use as reasonable these days (not low, not high, but reasonable).

[ATN69]

I love the sound of Diva but I did not buy it. I tried it but even though I have a i7-3.0GHz and 8Gb RAM, I could only accomplish 1 or 2 instances with Diva only before my CPU hit the roof. I did all optimizations I could in the DAW and audio settings as well. I even started to check if it would be possible to have Diva installed on another laptop, all alone, and somehow sync it with my main computer. I do own Zebra2, Hive and ACE. Bazille was also on the CPU limit for me. I guess I will try it again once I buy the latest 10 core i7 3.5GHz from Intel. I did not give up yet anyway ..for RePro I can already use it as it is, even though it is a bit hungry as well.

I love the sound of RePro. I used to own Korg MS20, Roland Juno 60, Korg Polysix, and many other synths. Based on my "feeling" I therefore choose filter 5 which is not so perfect in the high frequency area. If there is anything specific about analogue filters in hardware, it is the fact that they are imperfect. They have it's own personalities. Everything was built up with discrete components with all tolerance issues and stuff going on. I believe it's hard to mimic that in software. Filter 5 gave me that dirty, not perfect, sound that I personally call analogue sound

[Funkybot's Evil Twin]

I can force Diva into high CPU usage (50% in Studio One's CPU meter) if I use Divine mode, set the voices to 16 and stack to 4, jack up the resonance and filter FM, and play 4 to 6 note chords with a long release. In that regards, yeah, I guess Diva is intensive, but I'm usually playing it in Great quality, with 8 voices and little to no stacking, with less brutal filter settings. In those conditions, Diva's CPU usage will be around 10% while playing pads, which is excellent. I've created projects with a Diva bass, a lead (and a double), and 2 keys/pads tracks and I didn't have to bounce down until it came time do a mix with all the other tracks in the large project needing effects.

[kylie]

I have no idea if that filter actually sounds more analog than the others, but I can't help preferring type 1.

[ATN69]

I can force Diva into high CPU usage (50% in Studio One's CPU meter) if I use Divine mode, set the voices to 16 and stack to 4, jack up the resonance and filter FM, and play 4 to 6 note chords with a long release. In that regards, yeah, I guess Diva is intensive, but I'm usually playing it in Great quality, with 8 voices and little to no stacking, with less brutal filter settings. In those conditions, Diva's CPU usage will be around 10% while playing pads, which is excellent. I've created projects with a Diva bass, a lead (and a double), and 2 keys/pads tracks and I didn't have to bounce down until it came time do a mix with all the other tracks in the large project needing effects.

That's why I still wonder if I am missing something here?!!? ..I heard other people saying the same thing as you. It's not that Diva kill my CPU with every preset I tested. I also tested to make my own sounds. In FLS I can choose to run multicore threading and I can set buffer to allow slightly more latency just to give more buffer. I tried all that. I can make a whole song with 10 instances of Hive with all drums, fx and all, and still only reach maybe 50% CPU load. Anyway ..I don't give up that easily. I will find a way

[electro]

Since Diva's release we suffered stress from two directions: There was some from the side of the people who insisted it didn't sound analogue. Those were easily ignored. Then there were the people who thought it took too much CPU, and that wasn't warranted. This thread is about latter. It's about showing that implementing the same physical/analogue model sounds different if done with CPU saving shortcuts.

Very strange complaints. Diva does indeed sound 100% analog and somebody recently did a matched Oberheim OB-X patch bank proving it. As far as CPU complaints, multicore is working fine and composing in Standard or Great then exporting in Divine is just common sense.