Open Source DSP based hardware synth

[Zaphod (giancarlo)]

btw, virus ti is incredibly good. I don't think you'll replace it easily: I have tons of software which sounds tiny, aliased and with incredibly bad and boring patches. Virus ti would be great also as virtual instrument.
The same for uad, their models would be great also in the pure native realm. Dont be fooled by the hardware incarnation.

[whyterabbyt]

Why hardware? I sold a Virus TI a while ago and want to fill the void by building my own. They are very expensive to buy new and I want to build something as close as I can get to it, for the lowest cost, without having a team of engineers. Also some sort of proto dev board route, that would be adaptable to create guitar pedals/reverbs/filters etc.

If ten man-years of your time works out cheaper than the cost of a new Virus, then you feel free to do that. But the most salient part of aciddose's point was that any regular PC could achieve the same ends without the significant additional constraints and issues of a minority hardware platform.

I kinda see what you're saying. But i'm pretty sure you're wrong on just about every level,

Except he's not.

this is 2014, there's a raspi that's more powerful than the macs in around 2007.

Erm, no. The lowest 2007 iMac was running a 2Ghz Core2 Duo. The Raspberry Pi foundation estimate the Pi's horsepower at that of a 300Mhz Pentium II.
And yes, this is 2014. We run native stuff on systems much faster than 2007 Macs now.

Oh and the thing is your PC's not that powerful, it's not really making the most of multicores, and none of your programs or synths are either.

Im starting to get the impression you've drunk more than your share of the KoolAid.

And yeah you can get a credit card sized computer that's as good or better spec than your PC for a fraction of the price.

Feel free to give links to back that one up. I just bought a 4Ghz Haswell 4770K with 16Gb of RAM, and 768Gb of SSD. Accept challenge, please.

Affordable super computing at $100 is around now

You've never used a supercomputer, I suspect.

, but no-one knows how to use it or what to do with it or make the most out of the multicore/multithread processors.

And yet people have known what to do with supercomputers for decades. What has caused this miraculous failure of congnition?

This seems like a decent exploratory project that might help benefit a few computer users down the line.

If you had a realistic view of the capabilities, maybe. But you've got such a naively inflated perspective on how capable these things are, then I doubt it. Just for the record, what you're doing is pretty much the same thing as a lot of people did when they first realised there were such things as tablets; this whole 'its smaller and cheaper and it'll magically be faster and more special but only I have the insight that we can use it to replace everything' thing.
But despite the fact that there's some decently nifty DSP being done on modern tablets, its fantasy that its equivalent in power to a fully-loaded modern workstation. And its extreme fantasy squared that your $25 development boards will have caught up with tablet-level horsepower in the next three years.

The biggest problem computing faces is working in parallel, concurrent languages like Open CL, are under used by even the makers of Logic. We have great tech, but no-one actually really uses it.

Well, except for the people who do, of course.

Look after all is said and done, just be glad you're not as stupid as me who's gonna spend the next three years writing oscillators in various assembly languages!! Don't pity the fool

Seriously, enjoy yourself, and best of luck. But get some realistic perspective, willya? You're not preaching to a bunch of wide-eyed bunch of noobs who'll gasp in awe at every (innacurate!) claim of world-changing wooo you make, you're arguing with developers with years of first-hand experience in software development in exactly the areas you're getting wrong.

[antto]

well, i already hinted that you can simply avoid the whole hardware aspect, and code your actual dream-synth as a software plugin instead
you'll get faster results this way, and it's cheaper

but you're like.. uber excited like "woohoo, i got a fully equipped kitchen for very cheap! i can now make all the nice things like cookies, cakes, chicken nuggets, woah, the big guys are gonna go out of business! hold on to your hats. and this is good because others will benefit, i will uhm.. share the recipies.. yeah"
o-kay ;]

[Chopper]

well, i already hinted that you can simply avoid the whole hardware aspect, and code your actual dream-synth as a software plugin instead
you'll get faster results this way, and it's cheaper

but you're like.. uber excited like "woohoo, i got a fully equipped kitchen for very cheap! i can now make all the nice things like cookies, cakes, chicken nuggets, woah, the big guys are gonna go out of business! hold on to your hats. and this is good because others will benefit, i will uhm.. share the recipies.. yeah"
o-kay ;]

spot on...
to the OP... i understand you sold and now miss your VIRUS, but although i am not a fan, I aknowledge its power and flexibility, and there is a reason why it sells truckloads at its price... be realistic...

[raikard233]

Side note:
If it was so simple using cheap hardware to get the same results increasing revenues, didn't access guys already tought about implementing something like that? They also have R&D...
Think about RME that designs their own chips...

[UltraJv]

Why hardware though? You need to have a reason for hardware and as it is, you're better off with software. There are zero advantages to hardware unless you start off without a PC.

If you are sending the audio back through an audio interface and into a PC anyway, how canyou justify running the code on dedicated hardware when you have massively more powerful hardware inside the PC?

You can't.

We can justify analog electronics because the computational cost of producing a reasonable model for a $5 circuit is potentially going to take $1000 of computing hardware. We can build far more complex analog circuits than our hardware will ever be able to match and it will remain almost indefinitely far less expensive to do so.

The virus isn't an analogue piece of hardware, it's virtual analogue machine meaning there's a chip inside. It's a little tiny computer with an operating system and everything. It's a computer with GPIO, a processor, some DAC's.ADC's, and a butt tonne of code. And a few pots.

Why hardware? I sold a Virus TI a while ago and want to fill the void by building my own. They are very expensive to buy new and I want to build something as close as I can get to it, for the lowest cost, without having a team of engineers. Also some sort of proto dev board route, that would be adaptable to create guitar pedals/reverbs/filters etc.

I kinda see what you're saying. But i'm pretty sure you're wrong on just about every level, this is 2014, there's a raspi that's more powerful than the macs in around 2007. For £25, that's not bad, is that the computer you were referring to costing $1000? Oh and the thing is your PC's not that powerful, it's not really making the most of multicores, and none of your programs or synths are either. And yeah you can get a credit card sized computer that's as good or better spec than your PC for a fraction of the price.

Affordable super computing at $100 is around now, but no-one knows how to use it or what to do with it or make the most out of the multicore/multithread processors. This seems like a decent exploratory project that might help benefit a few computer users down the line.

The biggest problem computing faces is working in parallel, concurrent languages like Open CL, are under used by even the makers of Logic. We have great tech, but no-one actually really uses it.

Look after all is said and done, just be glad you're not as stupid as me who's gonna spend the next three years writing oscillators in various assembly languages!! Don't pity the fool

GPU or parallel processing dosnt work well for audio. Theres too much latency. The more of your posts I read the more its obvious that you havnt researched much as you tell us all hows the industry had missed a bunch of tricks. It has all been looked into and you just don't get it. You think any old oscillator/filter/Fx etc can be coded and it will be like a Virus. Not true, youll never do it. Its the special nature of coding that Access have used. They wont tell anyone how its done. Audio magic has always been hard to obtain, thats why there are so many EQs, filters etc but all sounding different.

Its the special technique of writing black magic audio code. You cant learn that, it requires lateral creative thinking. There have been a number of Virus look alikes that have the same front panel or controls but not the Virus sound. This is not a technical challenge but a creative one. If you don't know what makes a Moog special or a TB303 special or Supersaw special then you wont have a chance.

[Miles1981]

It's not only the latency, it's the fact that everything is dependent on the sample computed before, so it can't be parallized on a GPU, except if you can write as FIR filters and waveshapers.

[Chopper]

Why hardware though? You need to have a reason for hardware and as it is, you're better off with software. There are zero advantages to hardware unless you start off without a PC.

If you are sending the audio back through an audio interface and into a PC anyway, how canyou justify running the code on dedicated hardware when you have massively more powerful hardware inside the PC?

You can't.

We can justify analog electronics because the computational cost of producing a reasonable model for a $5 circuit is potentially going to take $1000 of computing hardware. We can build far more complex analog circuits than our hardware will ever be able to match and it will remain almost indefinitely far less expensive to do so.

The virus isn't an analogue piece of hardware, it's virtual analogue machine meaning there's a chip inside. It's a little tiny computer with an operating system and everything. It's a computer with GPIO, a processor, some DAC's.ADC's, and a butt tonne of code. And a few pots.

Why hardware? I sold a Virus TI a while ago and want to fill the void by building my own. They are very expensive to buy new and I want to build something as close as I can get to it, for the lowest cost, without having a team of engineers. Also some sort of proto dev board route, that would be adaptable to create guitar pedals/reverbs/filters etc.

I kinda see what you're saying. But i'm pretty sure you're wrong on just about every level, this is 2014, there's a raspi that's more powerful than the macs in around 2007. For £25, that's not bad, is that the computer you were referring to costing $1000? Oh and the thing is your PC's not that powerful, it's not really making the most of multicores, and none of your programs or synths are either. And yeah you can get a credit card sized computer that's as good or better spec than your PC for a fraction of the price.

Affordable super computing at $100 is around now, but no-one knows how to use it or what to do with it or make the most out of the multicore/multithread processors. This seems like a decent exploratory project that might help benefit a few computer users down the line.

The biggest problem computing faces is working in parallel, concurrent languages like Open CL, are under used by even the makers of Logic. We have great tech, but no-one actually really uses it.

Look after all is said and done, just be glad you're not as stupid as me who's gonna spend the next three years writing oscillators in various assembly languages!! Don't pity the fool

look, without wanting to be harsh, you might want to listen to what aciddose is telling you there, and investigating who he is and what the guy did/knows/is capable of before replying that kind of nonsense....

[VariKusBrainZ]

Remember when Williamk (wusik) was going I write his own OS dedicated to music production........

[FabienTDR]

^^ +1

b&t your last post contains so much nonsense, it isn't even worth going into details.

A single core or a modern CPU alone (= non parallelisation at all) is easily a hundred thousand times faster than the chips found inside a Virus. The problem is, all the nerdy tech you are waiting for has absolutely nothing to do with sound. The 20 year old Viruses essentially prove it. I think you are looking at the wrong places. And in particular, I recommend you to put all your half-wisdom to practise, it will help you sorting some things out.

"DIY" signal processing has far more important problems than GPU or full CPU support. Violation of the sampling theorem is probably the main bottle-neck. In order to solves this problem inside a something as complex as a synthesizer, serious brain work is needed. But it has nothing to do with secret coding techniques (lol) or factors only important for a linux user group meeting.

Reading your text, I had to think about:
https://www.youtube.com/watch?v=-MM0Z_jCrGo

[UltraJv]

Is is time for the cat pics yet?

[WilliamK]

Well, I have the XMOS kit here, and is a nightmare to program. For threads to talk to each other you need some fancy stuff like channels and connections, as the whole thing is real-time. Once you get how it works, is powerful, don't get me wrong, is just super fast. But, it takes a while to understand things. So, when we enter the open-source DIY market, we have to remember that most musicians will want to focus on music and want to make easy hacks.

Check out my own project, which is Arduino Based, and is a great and simple one: www.Beat707.com

Now, yes, several years ago I was working on my own OS, it was the Dash OS, for Dash Synthesis. It was Linux based, but too much work had to be done before anything would be useful. I lost funding for it and let it go. Now we have similar stuff, but all closed-source: Korg Kronos is a great example, as it is a linux Atom based machine. Muse Receptor is another one. So it is possible, but it takes a lot of work to be done.

Now, XMOS could change things, as they have great chips, but they are hard to program.

Hardware open-source is fun, so I wonder how this will continue. I see a lot of people here bashing the whole idea, specially since home computers are so powerful. But, keep in mind that some people (Me) loves to play around with the limits of some things. On my case, I love to see how much I can do with a simple 8 bit 16Mhz processor such as the Arduino Uno has. But that's me... and a bunch of other people too...

I have seen the latest Pulsar/Scope XITE boards, but they costs nearly 4k Euros, which is insane. But them, zero-latency... that's a big deal. I finally got my computer a decend usb card that has 2ms latency, but it clicks at random times, and it was hard to get this in Brazil. You guys need to remember that Europe and the USA is not the whole world. For people like me, in countries like Brazil, simple stuff can cost 2x to 3x more, due to import taxes and what-nots... so, been able to buy something in parts is one solution.

EG: This month I ordered the PCB and the Microprocessor. Next month I will get the case. 4 months later I finish up building a device that I couldn't purchase from overseas assembled... and I can also tweak the code for some ideas I had... See?

Anyway, just thinking out loud.

[b&t]

damn just lost everything I wrote

essentially, I back down. I'm still curious how it's achieved.

[b&t]

..., we have to remember that most musicians will want to focus on music and want to make easy hacks.

Check out my own project, which is Arduino Based, and is a great and simple one: http://www.Beat707.com

Now, XMOS could change things, as they have great chips, but they are hard to program.

Hardware open-source is fun, so I wonder how this will continue. I see a lot of people here bashing the whole idea, specially since home computers are so powerful. But, keep in mind that some people (Me) loves to play around with the limits of some things. On my case, I love to see how much I can do with a simple 8 bit 16Mhz processor such as the Arduino Uno has. But that's me... and a bunch of other people too...

Anyway, just thinking out loud.

Yeah starting to think the xmos approach of open source and the whole let other figure out what to do with it, is genius but flawed. Great way to get free man hours of dev by open source. They hope someone makes a cool project so they sell but tonnes like the arduino/raspi kinda way but it's exponentially more difficult to program.

The affordable super computers are here: http://www.parallella.org

[b&t]

The $30 synth kit is genius and the Cortex based drum machine, the monome is great example of open source awesomeness too, and the shruthi synth is wicked too. If there's a will there's a way, it's just finding the right way, I was just curious how one would go about it about through a different route. Explore whether there's a potential to solve some problems in a different way based on the tech we have at our disposal. I was just curious.

Thank you to everyone involved, it has been a fun conversation, learned a lot, gained perspective, and glad you all feel so passionately on the subject.

I realise that it's quite a challenge down the suggested route, and probably best to just get one of those $30 kits and make some squiggles!

EDIT: side note, here's an interesting vid I watched last night on thinking in parallel: http://www.infoq.com/presentations/Thin ... rogramming