u-he rePro in the works

[fluffy_little_something]

In that case he might as well have kept his new project secret and presented his finished emulation/product one day...

[nevernamed]

And on the Diva hardware thing what might make sense is partnering up with a hardware company (risk management)?

I think it makes sense to forget the idea and keep making cool software synths like they are obviously good at and can be successful doing. That door is wide open, full of creative potential and has little risk and lots of reward and the iron is hot so to speak.

There is already 3-4 years worth of projects wanting to be completed. Down the road, who knows. Keep making great stuff, and some hardware manufacturer will approach u-he.

There are ways to mitigate the risk though. Kickstarter for example. And there is nothing quite like lots of hands on control. Some of us have started moving away from the computer a bit. I've gotten rid of all of my processing plugins pretty much. I bought a Harrison 950MX instead and a whole shitload of outboard (Neve 1073s, Pacificas, 1176s, Retro STAs etc etc) and some great microphones from Bock and AEA. I started ITB but now the computer has become sort of like a tape machine for me. I keep it around because I like Pro Tools editing and I have a few synths from U-he. Going forward though I see computers disappearing from pre-production. For me I mean. Pro Tools will still kick around for me because that is is where everything will come together during the production stage but other than that I see no role for the computer.

I'd like to get rid of my old analogues because I'm tired of servicing them and I think they'll eventually die anyway hence the U-he in hardware plea.

[Funkybot's Evil Twin]

Nevernamed, I hear what you're saying, but what would that cost? I'd be in for DIVA [quality] in hardware form for $1,500 but I'm not sure if that's realistic. I'm talking 4 octave keyboard, onboard CPU/DSP, enough knobs, faders, and buttons to make things comfortable, and the ability for it to function as a controller for other things. If the cost ended up being closer to $3,000, which it would probably have to be, then I'd lose interest (could get a Prophet 6 or the new Oberhiem for that).

[fmr]

CPU Power will not grow the next years. Maybe for some highend desktop PCs but more and more people will use only laptops and Surface like tablets. Therefore Software witch needs low ressource requiremenst will win the race.

Yes, and we will see the same panorama we are now confronted in music , when we passed from the Ultra Hi-Fi paradigm to the MP3 paradigm as a consumer product...

But there will still be a demand for more powerful products, and people buying real powerful computers to get the job done.

[TheNickC]

If I were a professional, I would not care if it cost 1000 or 1500 or even 2000 bucks, frankly
Makes me wonder how many people using very demanding synths actually are professionals. I assume the majority are not and use them on mediocre computers.

I'm a hobbyist and I might consider spending that on a CPU to run as much Bazille as I want. Yes you can ostensibly sell more units making something super efficient and going after the iphone market, but there's room for uncompromised excellence to exist too. Desktop computers may become more niche but they aren't going away, having a powerful computer with a big screen and speakers and other gear, there's merit to that as well and not just for audio.

[Urs]

Slightly more on topic - the future projects Urs has talked about rely upon faster processing. Presumably, he believes that Moore's law will hold for a few years?

I don't assume anything . I just want to move on to the next step in analogue modeling. This might involve more complex models than what we have in Diva, hence more CPU power. So I spend some time not just coming up with these models but also with different approaches on how to implement them.

For instance, in Diva there's always only a choice of one algorithm, determined by the quality setting. There are however different ways to achieve the same (measurable) quality:

- spend more time on the initial estimate, then hope for less iterations
- or spend less time on initial estimate at risk of more iterations
- use approximations to divison, square root and tanh() but risk more iterations
- or use precise maths in hope of less iterations
- use multiple convergence checks in hope for "half iterations", but add extra cycles
- or use just one check to save those cycles, but always do "full iterations"

So here are 3 pairs of opposite that may or may not lead to faster processing. The filter strcuture, filter parameters as well as input material determine which of the cases and combinations thereof are faster. Which bears a new set of cases:

- dynamically switch algorithms per filter for speed, but risk cache pollution
- dynamically switch algorithms per process in hope for an overall speed up, but spend more time finding it
- stick to a universal algorithm that's empirically the fastest, benchmarked by all known sound banks

Which may depend on processor, number of cores, cache size... but those are all questions that come way after "how accurate is accurate enough?"

[Urs]

In that case he might as well have kept his new project secret and presented his finished emulation/product one day...

As I wrote in the announcement post, we wanted to secure the name/trademark before somebody else grabs it.

[Urs]

What about CUDA?

Not yet suited for audio filter processing. (But good for convolution)

[studiowaveform]

What about CUDA?

Not yet suited for audio filter processing. (But good for convolution)

Nvidia need to go out clubbing more

[david.beholder]

Nvidia need to go out clubbing more

To underground clubs not to commercial edm-parties, otherwise we would end up in first cuda based supersaw emulator with armin tiesto branding.

[nevernamed]

Nevernamed, I hear what you're saying, but what would that cost? I'd be in for DIVA [quality] in hardware form for $1,500 but I'm not sure if that's realistic. I'm talking 4 octave keyboard, onboard CPU/DSP, enough knobs, faders, and buttons to make things comfortable, and the ability for it to function as a controller for other things. If the cost ended up being closer to $3,000, which it would probably have to be, then I'd lose interest (could get a Prophet 6 or the new Oberhiem for that).

Those are all sensible points mate. I don't know how much it would cost. The Waldorf Q is in the $1500-to-$2000 range on the used market. Some of the analogues are closer to 3k like you pointed out. I think u-he in hardware would offer a bit more than say a Prophet 6 or Oberheim. It can cover a lot of of the analogue style sound but it can also do digital stuff that the analogues can't do. I bring up the Q because it's a VA, wavetables, knobby interface, lots of modulation, a range of filters and it sounds great so perhaps it's better to think about it along those lines. Then there is whole Roland plugout thing too. That concept is interesting in that you can upload new features/models over time. Anyway I'd rather have that sort of thing than the new Prophets and Oberheims and I think it would be worth about the same.

[nevernamed]

Nevernamed, I hear what you're saying, but what would that cost? I'd be in for DIVA [quality] in hardware form for $1,500 but I'm not sure if that's realistic. I'm talking 4 octave keyboard, onboard CPU/DSP, enough knobs, faders, and buttons to make things comfortable, and the ability for it to function as a controller for other things. If the cost ended up being closer to $3,000, which it would probably have to be, then I'd lose interest (could get a Prophet 6 or the new Oberhiem for that).

Those are all sensible points mate. I don't know how much it would cost. The Waldorf Q is in the $1500-to-$2000 range on the used market. Some of the analogues are closer to 3k like you pointed out. I think u-he in hardware would offer a bit more than say a Prophet 6 or Oberheim. It can cover a lot of of the analogue style sound but it can also do digital stuff that the analogues can't do. I bring up the Q because it's a VA, wavetables, knobby interface, lots of modulation, a range of filters and it sounds great so perhaps it's better to think about it along those lines. Then there is whole Roland plugout thing too. That concept is interesting in that you can upload new features/models over time. Anyway I'd rather have that sort of thing than the new Prophets and Oberheims and I think it would be worth about the same.

[pdxindy]

Nevernamed, I hear what you're saying, but what would that cost? I'd be in for DIVA [quality] in hardware form for $1,500 but I'm not sure if that's realistic. I'm talking 4 octave keyboard, onboard CPU/DSP, enough knobs, faders, and buttons to make things comfortable, and the ability for it to function as a controller for other things. If the cost ended up being closer to $3,000, which it would probably have to be, then I'd lose interest (could get a Prophet 6 or the new Oberhiem for that).

Those are all sensible points mate. I don't know how much it would cost. The Waldorf Q is in the $1500-to-$2000 range on the used market. Some of the analogues are closer to 3k like you pointed out. I think u-he in hardware would offer a bit more than say a Prophet 6 or Oberheim. It can cover a lot of of the analogue style sound but it can also do digital stuff that the analogues can't do. I bring up the Q because it's a VA, wavetables, knobby interface, lots of modulation, a range of filters and it sounds great so perhaps it's better to think about it along those lines. Then there is whole Roland plugout thing too. That concept is interesting in that you can upload new features/models over time. Anyway I'd rather have that sort of thing than the new Prophets and Oberheims and I think it would be worth about the same.

Besides not wanting U-he to spend lots of time on things other than new and existing soft synths... I think now is a bad time for a hardware version.

I'm unlikely to buy a hardware Diva. I already have Diva. While ergonomics could be better in a dedicated controller, I'm not exactly unhappy with how I control it now. That level of cost would be too much for what is essentially a bit of convenience.

If you make the hardware version better than the soft one to differentiate it and justify the high cost, you annoy the software users.

Now, if U-he came out with a hardware synth that had Diva, Zebra, Bazille, ACE and Hive in it... and could be upgraded with new synths when they arrive... and a place to dock an iPad to function as a screen for touch editing the diverse controls of the different synths... that at least sounds tempting and does not easily obsolete itself.

[nevernamed]

All good points. And it's exactly what I was saying. The hardware concept needn't be limited to just VA stuff. They could include all of their synths of course and flirt with something like the plugout architecture as an obsolescence control. It needn't happen tomorrow of course.

[pdxindy]

All good points. And it's exactly what I was saying. The hardware concept needn't be limited to just VA stuff. They could include all of their synths of course and flirt with something like the plugout architecture as an obsolescence control. It needn't happen tomorrow of course.

I'd rather it not happen for at least 3 years...