Open Source DSP based hardware synth

[b&t]

Just got back from a OpenSource Hardware conference/hackers festival, and pretty inspired by some of the talks and workshops from XMOS, in particular doing DSP on a board that costs £10 and an audio slice that plugs in for £25.

So to make a DSP based hardware synth, we could get there for under £50 and all with open source pre-built proto boards, so the boards probably actually cost £2 and £7 for them to make.

I would like this project to be open source, as the boards are open source, and I think it opens up more possibilities for learning and engagement, and potentially unparalleled level of development due to low cost of boards and open nature.

So there's potential for you guys to make serious money taking your lovely C++ code and putting it in hardware*. It will probably be more fun than chucking it onto something you don't have to care about the GUI

*(After you know, a few years down the line and we've done all the hard work, and there's a nice youtube tutorial!!)

Also we have a new paradigm opening up with max/msp being able to export low level C++ DSP code straight from gen~, so the guts could be prototyped in a familiar graphical environment. And an engagement from an audience interested in interfacing hardware and software could be a giant bonus for dev. I swear a synth is the very simplest thing that will come out of this! We could essentially build an octatrak for under £50, with the same guts, this is a new age of tech

If you guys have any thoughts or contributions on the subject matter, please feel free to chime in. Would really like to know whether this could be a decent avenue of exploration.

Please no posts on "what about knobs and teh shiny case". I just want a legitimate interest level from the seeming "authority on music DSP", please no attacks from companies

[cocoazenith]

Just got back from a OpenSource Hardware conference/hackers festival, and pretty inspired by some of the talks and workshops from XMOS, in particular doing DSP on a board that costs £10 and an audio slice that plugs in for £25.

So to make a DSP based hardware synth, we could get there for under £50 and all with open source pre-built proto boards, so the boards probably actually cost £2 and £7 for them to make.

I swear a synth is the very simplest thing that will come out of this! We could essentially build an octatrak for under £50, with the same guts, this is a new age of tech

Amazing news. I'm not a developer, but a graphic artist, but I'm so looking forward to this. And I can't help to fantasize about the case design that's gonna follow after the "scientific" part becomes a reality.

[Chopper]

We could essentially build an octatrak for under £50, with the same guts, this is a new age of tech

pretty unrealistic, if you are aware of what is inside an Octatrak's guts...
Not trying to be negative here, don't get me wrong...

[b&t]

pretty unrealistic*, if you are aware of what is inside an Octatrak's guts...
Not trying to be negative here, don't get me wrong...

*pretty idealistic, not unrealistic. They are probably going a different route like balckfin's or something similar. XMOS is concurrent lowest latency, 192khz, 24bit, crazy low price, idealistic, not unrealistic, we have more spirit in open source and win in numbers less you forget.

A mono/poly synth could be easy, especially piggy backing off the Openframeworks in C++, it's like the processing of DSP. But also gen~ and writing upsampled oscillators and freak man we could even use neural nets to make recursive filters and challenge these people at UAD, for a fraction of the price.

We have the potential to make Access/UAD/Nord, basically go out of business, by development from the open source community. All these £1000 devices TI/NordLead/octatrack etc as built on boards you can put in your hands for under £50 at present.

We as developers now have access to reverse engineer any of these devices ourselves and no go out of pocket developing.

I guy I sat with has an early chip board from 1989/1990 or something he got out to show the speaker as a relic, paid £4,000 for the privilege of owning one back them, I got given one as I entered yesterday for free, that was, let's just say quite a lot better. And £25 for the audio slice is not bad, so I'm only out £25, and for others proto-ing a DSP synth like the Virus or Nord it's a pretty great entry level price wouldn't you agree? You could even use the prebuilt boards, a case you build, times all by 5, get £250, and still make loads of profit.

Glitch machines, live granulation, we can do better than the Octatrak actually

[raikard233]

Well, dunno if you ever had experience in making hardware stuff, thinking about hardware layout, ergonomy, quantity prices and so on about components.
The prices of some hardware include knowledge, reasearch and development time, design/ergonomy research and engineering, electrical safety requirements, different world standards etc.. all details to be considered and studied at project start... and so on.
There are so much variables inside a commercial project that can raise the costs in a snap, especially if you make your product in limited quantity, by hand, in house, in a "western" country.

Before saying that we can make out of business certain manufacturers i would like to make some more realistic check, a lot of people don't do DIY, they want realized and finished products, they want assistance and don't want to put hands inside something that they don't know.
Some users want an instrument ready to be used and want assistance if it's not working/having problems.

It's not all so easy and simple as you might think.

I've created a complex 4 channels Traktor controller some years ago, almost industrial standard, with aluminium case.
But the costs are so high that would need a medium investment for a series of at least 500 units to make the price raise down and make really affordable using quality components.
Also making it as a DIY kit (nothing really complex as it's a midi controller) will not cost less than 1000/1200€.

Making an Octatrack for 50£ including case, pcb, code, eventual assistance, time spent on building and testing... you can't say that's realistic. Unless you live on the fat of the land.

My 2 cents, of course not to low down the enthusiasm.

[b&t]

Well, dunno if you ever had experience in making hardware stuff, thinking about hardware layout, ergonomy, quantity prices and so on about components.
The prices of some hardware include knowledge, reasearch and development time, design/ergonomy research and engineering, electrical safety requirements, different world standards etc.. all details to be considered and studied at project start... and so on.
There are so much variables inside a commercial project that can raise the costs in a snap, especially if you make your product in limited quantity, by hand, in house, in a "western" country.

Before saying that we can make out of business certain manufacturers i would like to make some more realistic check, a lot of people don't do DIY, they want realized and finished products, they want assistance and don't want to put hands inside something that they don't know.
Some users want an instrument ready to be used and want assistance if it's not working/having problems.

It's not all so easy and simple as you might think.

I've created a complex 4 channels Traktor controller some years ago, almost industrial standard, with aluminium case.
But the costs are so high that would need a medium investment for a series of at least 500 units to make the price raise down and make really affordable using quality components.
Also making it as a DIY kit (nothing really complex as it's a midi controller) will not cost less than 1000/1200€.

Making an Octatrack for 50£ including case, pcb, code, eventual assistance, time spent on building and testing... you can't say that's realistic. Unless you live on the fat of the land.

My 2 cents, of course not to low down the enthusiasm.

See first post, still not talking about a case or knobs EDIT(nor commercial product, this is open source). Pure prebuilt proto boards, that are fast enough to do DSP, and easy enough to use so you're not talking in assembly or machine code.

They do sell a DJ kit, for some amount, but seems sensible to me going Raspberry pie > Xmos Start kit > XMos audio slice. £30/£10/£25, so over £50, but that's for the guts of a pretty competitive piece of hardware.

Idealism and optimism my friend, they can do it, then so can we, and here's a potentially cheap hack through for prototyping using the latest technology, a new language XC, for some concurrent tricks best suited for DSP.

I'm just wondering if I can make a super hack through and use someone else's C++ frameworks, like Maximilian, but reckon this stuff would give Urs a total geek boner, and would so buy a hardware version of DIVA right away! That dude alone could put Access out of business, so I'm really not being too idealistic, in fact I'm the realist, as this is the (con)current reality!!

I really don't mind people's pessimism, it's not going to affect me.

[b&t]

Look it's not just me: http://www.rebeltech.org/shop/ - link protected apparently, search OWL open source effects box

that guys is doing an open source hardware effects box you can program in C++, £250, sound familiar? So £50 to make including case. When I say make, I mean manufactured, he got crowdfunded £30k or so to do that, but I'm suggesting dev's can proto a similar device as what he's built, that's better for less.

[raikard233]

The project in itself it's interesting about open source code etc (let's start digital vs analog flame ), but the cost it's always something in the standard, i don't see nothing so revolutionary to bring down other companies.
Then we are talking about a finished product, with 4 pots and a switch with 4 jacks for 250£, of course customizable, but always 250£, and if you want more instances, welcome to the hardware paradigma.
Think about when you have to deal with plenty of button, pots and encoders, mutliple outputs.

[b&t]

The project in itself it's interesting about open source code etc (let's start digital vs analog flame ), but the cost it's always something in the standard, i don't see nothing so revolutionary to bring down other companies.
Then we are talking about a finished product, with 4 pots and a switch with 4 jacks for 250£, of course customizable, but always 250£, and if you want more instances, welcome to the hardware paradigma.
Think about when you have to deal with plenty of button, pots and encoders, mutliple outputs.

No I'm not talking about a finished product ffs. I'm talking about prototyping a synth, with a few pre built boards, and please don't tell me pots are expensive, I have an abundance knocking around, so they are always free to me!

I want a synth anyone can put together, but the hard stuff, the computer/DSP/DAC's are all done and out the way for £50 (Knobs not included). That's with pre built boards you plug together. Pieces you can buy for next to nothing right now. Is it really that hard to push such a simple idea.

It would put these guys out of business if everyone had the natural curiosity to make things themselves.

Also see first post can we not talk about USER INTERFACES please. Very much a bring your own to the party... assuming this ever gets out of the "what you gonna do about the knobs" stage.

[aciddose]

I've done a synthesizer for $7 using a z8e chip. You don't really need anti-aliasing as you can just generate the signals at a rate like 5mhz - all you need is to use integer divisions of your master clock and the reflected aliases line up on existing harmonics. At low rates (44.1k) tuning is an issue, but this disappears at high rates.

The problem is: what is the point? It is both easier and cheaper to do in software. It also provides potential for a far better interface, makes development extremely easy, upgrades/improvements easy and the quality of the product can be far higher than any DSP chip could provide. Remember we have $1000 or more of hardware in a modern DAW. Sure, match that with a DSP and you can approximate it, but you need to look at everything in combination, not just memory and processing power.

If you want to do a prototype I'd do that in software first then once you have something you are satisfied by, port that to the DSP processor.

[raikard233]

The project in itself it's interesting about open source code etc (let's start digital vs analog flame ), but the cost it's always something in the standard, i don't see nothing so revolutionary to bring down other companies.
Then we are talking about a finished product, with 4 pots and a switch with 4 jacks for 250£, of course customizable, but always 250£, and if you want more instances, welcome to the hardware paradigma.
Think about when you have to deal with plenty of button, pots and encoders, mutliple outputs.

No I'm not talking about a finished product ffs.

To knock out these guys you have to talk about finished products. Otherwise you're doing apples vs oranges.

I'm talking about prototyping a synth, with a few pre built boards, and please don't tell me pots are expensive, I have an abundance knocking around, so they are always free to me!

To you, not to everyone. I have a lot around too, but i paid for them.
Pre built boards cost less if done in quantity, not if self built or in 1-5 pieces (if we talk about DIY).

I want a synth anyone can put together, but the hard stuff, the computer/DSP/DAC's are all done and out the way for £50 (Knobs not included). That's with pre built boards you plug together. Pieces you can buy for next to nothing right now. Is it really that hard to push such a simple idea.

You can do it, but the funniest thing about hardware synths is knob twisting over a well designed front-panel.

It would put these guys out of business if everyone had the natural curiosity to make things themselves.

I have the curiosity to make things myself, that's why i've built that Traktor midi controller to spend less and have more controls than the other controllers available into the market at that time (2011).
But at the end i've realized that some stuff start to being convenient only if done at industrial standards.
If you do DIY, don't expect less expense, you gain a lot of knowledge of course, but not so much when talking about wallet.

Also see first post can we not talk about USER INTERFACES please. Very much a bring your own to the party... assuming this ever gets out of the "what you gonna do about the knobs" stage.

Ok, no problem to me, i had only the like to bring my experience about this stuff.
I'll stop here, but only one thing i would like to remark: when you talk to push out of business guys (Access, Elektron, Nord), you've to make a 1:1 comparison. So bring to the party knobs, layouts, manuals and support.
Otherwise you're into the Apples VS Oranges field.

Peace.

[antto]

you could take an IDE, and a compiler, and make your own synths, and then knock off all the big guys out of business

[b&t]

I've done a synthesizer for $7 using a z8e chip. ...

The problem is: what is the point? It is both easier and cheaper to do in software. It also provides potential for a far better interface, makes development extremely easy, upgrades/improvements easy and the quality of the product can be far higher than any DSP chip could provide. Remember we have $1000 or more of hardware in a modern DAW. Sure, match that with a DSP and you can approximate it, but you need to look at everything in combination, not just memory and processing power.

If you want to do a prototype I'd do that in software first then once you have something you are satisfied by, port that to the DSP processor.

yeah this XMOS board isn't just a chip, it's a chip, but more like a multi thread co-processor for say a raspberry pie.

What intrigues me, is the whole of it could be done in software, as you can talk to XMOS in C/C++, so it's ideal for developers to take their code and put it on a device.

Oh also the funny thing I learned from the conference, from one the lady's that designed the ARM chip, said that processors haven't been getting any faster since 2006 due to Amdahl's law, which actually is law as opposed to Moore's one, see failing of Pentium four i think it was. Software needs to be written in a concurrent languages to make the most of what these processors can do.

*And @raikard2333, sorry i should have never said anything about competitors or competition or knocking these guys out the way. This is just for fun, and just exploring some tricks I've become aware of, i'm just sharing some ideas with the forum, sorry for invoking capitalism and manufacturing, merely excited and inspired.

[b&t]

you could take an IDE, and a compiler, and make your own synths, and then knock off all the big guys out of business

sarcasm, doesn't come across well on the internet remember.

Idealism and being up-todate with the latest technology has and always will knock the big guys out of business.

IDE, Compiler, Audio Slice, Raspberry Pie, XMOS starter kit, and few pots and potentially make a software synth standalone.

[raikard233]

Look at this: http://www.sonic-potions.com/lxr
That's a really nice project, would be possibile to make it using these chips that you're talking about?

It's always welcome to me DIY stuff, don't get me wrong