 |
|
You are not logged in! F.A.Q Log in Register |
 |
|
 |
 Now online (2)w M w steve mcqueen ...and 166 guests Last 5 registered Oplandisks nothingstar N_loop yipe foxtrotromeo Browse members... |
 |
Members 8025Messages 2613420 Today 12 Topics 127500 |
|
|||||||||
| |||||||||
|
|
I was watching a having a discussion tonight with my brother tonight about fourier transforms and how they are used for digital encoding, etc, and i was discussing in the context of sound and how every conceivable soundwave can be constructed by compiling an infinite number of sinewaves with different frequencies and no amplitude. Well i was thinking of if you had a finite series of sinewaves and you could vary their frequency coefficients on the fly how you could create really wild and interesting sounds. Is this in essence what digital synthesizers do in a finite sense, as i imagines its very computationally expensive to do things in real time, i suppose thats what wavetable synths do. does anyone know what im on about, has anyone go a concept of what im getting at? |
||||||||
|
|
|||||||||
| |||||||||
|
|
sure that's bruv |
||||||||
|
|
|||||||||
| |||||||||
|
|
^ syeah your right what i meant with the wavetable is a list of variables stored for manipulation, like a look up table, like a very simple fourier transform, im not sure im right about that, actually just checked the wiki article they are different methods of achieving the same thing. |
||||||||
|
|
|||||||||
| |||||||||
|
|
Kind of yes, in both cases you're morphing between spectra, the difference is how you model them, whether as waveshapes or a graph of partials and their amplitudes. IMO U-he Zebra is the best for this kind of thing even if the default GUI looks a little ropey in 2016 terms. |
||||||||
|
|
|||||||||
| |||||||||
|
|
cheers, yeah sorry i only partially understand what im going on about, that zebra thing looks like what i was initially trying to describe |
||||||||
|
|
|||||||||
| |||||||||
|
|
discrete or continuous methods of generating a waveform |
||||||||
|
|
|||||||||
| |||||||||
|
|
it is 'quite interesting' - eg for analysis - the longer the time you look at the more frequencies you can see but less information about WHEN they occurred - shorter time the opposite... makes you think about how much the two things are literally just different ways of looking at the 'same "thing"' |
||||||||
|
|
|||||||||
| |||||||||
|
|
like, if you play a sine wave who's cycle lasts a year, and you analyse the first weeks worth and try to build that up from sine waves lasting a week... |
||||||||
|
|
|||||||||
| |||||||||
|
|
How do you know how long the wave will last if we've seen a just little bit |
||||||||
|
|
|||||||||
| |||||||||
|
|
best u ignore most of that, cant explain for toffee |
||||||||
|
|
|||||||||
| |||||||||
|
|
been loads of good spectral synthesis things like MetaSynth and ZynAddSubFx etc. but they don't get much of a mention in books I've seen...seems like we've invented all the synthesis techniques possible! bonkers. |
||||||||
|
|
|||||||||
| |||||||||
|
|
yeah all the fundamentals have been covered i agree, its amazing to think the concept of fourier transforms existed in pure maths hundreds of years before it had practical applications in the modern world, blows me noggin, imaging a program that could generate radically different waveforms quickly and programatically without you having to change any parameters, you could observed the creations and save preferred settings, there is so many variables im sure there are still many many sounds unexpected out there. |
||||||||
|
|
|||||||||
| |||||||||
|
|
imagine a hardware synth where you can have almost vast banks of waveforms interacting, but generating something that approximates usable sounds that complement each other rather than generating something that sounds like white noise |
||||||||
|
|
|||||||||
| |||||||||
|
|
suppose granular synthesis is the closest to what im talking about but thats individual grains, im talking about something that has so many oscillators you couldnt manually change any of the varibles cos there is so many of them, but if you had a system behind it that handles maths, you could have preset algorithms to handle everything for you, so you could change any of the variables in the algorithm and it would change everything, you could have it as attuend or chaotic as you wanted |
||||||||
|
|
|||||||||
| |||||||||
|
|
what if i should want five? |
||||||||
|
|
|||||||||
| |||||||||
|
|
yes we could do five your the technical guy arent you Epic we would need a super computer for that, or an array of raspberry pi's to spread the load |
||||||||
|
|
|||||||||
| |||||||||
|
|
sample and phase accurate sync between computers would be required else you would have phase distortion, cancellation or reinforcement of freqys. you should check out max msp you can achieve what you like with that. |
||||||||
|
|
|||||||||
| |||||||||
|
|
yes phase cancellations would deffo be something i would envisage if i was rendering in real time performance, I think if the system was programmed correctly that wouldnt be an issue as id get it to render sound only when calculations were complete, yes max msp looks very good, i might see what i can accomplish with something like supercollider for a laugh, get my classes running on separate threads |
||||||||
|
|
|||||||||
| |||||||||
|
|
max 6 opens doors to new worlds of creativity. |
||||||||
|
|
|||||||||
| |||||||||
|
|
ah you know already, I posted as you posted. |
||||||||
|
|
|||||||||
| |||||||||
|
|
max is great because there are allot of tutorials. |
||||||||
|
|
|||||||||
| |||||||||
|
|
thanks for link, i havent checked it out for ages |
||||||||
|
|
|||||||||
| |||||||||
|
|
the flow chart reminds me of stuff like reaktor, but i suppose behind that are you own classes rather than pre configured objects |
||||||||
|
|
|||||||||
| |||||||||
|
|
you can make your own objects but there are hundreds already made and very efficient. you can change how they behave too just join them up with wires/cords. lol its fun. random frequencies |
||||||||
|
|
|||||||||
| |||||||||
|
|
so much fun creating funny noises, first thing i did on supercollider was create a waterfall sound, low pass filter on white noise, can remember what else i did |
||||||||
|
|
|||||||||
| |||||||||
|
|
i wouldnt necessarily use supercollider to make music but to make really cool samples its ace |
||||||||
|
|
|||||||||
| |||||||||
|
|
You can make so much stuff these days with these types of programs. i love using objects and cords. |
||||||||
|
|
|||||||||
| |||||||||
|
|
yes, its a more than legitimate alternative, i would do that if i had the cash, you can get the sounds you want quicker |
||||||||
|
|
|||||||||
| |||||||||
|
|
https://www.youtube.com/watch?v=GtAMYafil2Q |
||||||||
|
|
|||||||||
| |||||||||
|
|
imagine if we could genetically modify ourselves to be human synths, we could think waveforms, and have midi ports under our armpits |
||||||||
|
|
|||||||||
| |||||||||
|
|
That's an interesting idea, the year long sine wave. Got me thinking... You could measure the rate of change of phase over the two week sampling period, and compute the frequency with good temporaral resolution because the rate of change of phase is constant for a sinusoid. That's great but it's really a special case - other waveforms wouldn't work. e.g. how would you do a square wave when it is off for 26 weeks and you only measure two weeks of 'off'. There are lots of interesting fourier techniques and things to wrap your head around, it really is amazing that it was developed mathematically so long before we could create DSP chips and properly utilise it. |
||||||||
|
|
|||||||||
| |||||||||
|
|
^ does phase shift occur with any sine source, is there nothing that doesn't eventually drift? is that just the limitation of electronic? |
||||||||
|
|
|||||||||
| |||||||||
|
|
Frequency is rate of change of phase. A sinusoid is a single frequency so a constant rate of phase change. Any distortion on the sinusoid affects the phase. In my imagination I can create a perfect environment where can ignore friction, etc. Imagine a wheel rolling past a long white wall. The wheel has a pen at one position on its perimeter. As the wheel rolls past the wall the pen plots a sinusoid on the wall. The wheel rolls at a constant speed because there is no friction. Constant speed also means constant frequency, so if it takes 1s to do a full turn then it is 1Hz. Rate of change of phase is 360 deg per second. By inspecting the line drawn on the wall after 0.1s it is easy to extrapolate to the end point because the curve of the line is constant, i.e. it forms a circle... it's just in this case the wheel rolls forward so the circle never closes. N.B. If we take the same wheel and put it on an axle it is clear that the pen plots a circle rather than a sinusoid. Obviously in the real world there is friction so it doesn't work out so well. Now if there was a year long sinusoid it could be very accurate. We could use an atomic clock to create 'ticks' that counted up. Each count indexed a memory location that stored a sample of our sinusoid. The sample read out of memory could be passed to a dithered ADC, where we can easily filter the dither with a low pass response without affecting the 1yr sine but smoothing out the changes. Everything would be held at a constant temperature. All voltages would be heavily regulated and filtered. All vibrations heavily damped. Or maybe the sinusoid is generated from some piece of junk that drifts all over the shop like an early Moog. Oh and perhaps it is distorted so the sinusoid is a bit wonky. The universe drifts so I suppose everything depends on where you are observing it from. I used to work with v. precise equipment and in the end it came down to loads of statistics and some philosophy about such things. Quite interesting. |
||||||||
|
|
|||||||||
| |||||||||
|
|
^ thanks for the explanation, i was thinking about atomic clocks before yes everything drifts from a relativistic sense, |
||||||||
|
|
|||||||||
| |||||||||
|
|
you don’t have to worry about this as long as all your experiment shares the same clock source. |
||||||||
|
|
|||||||||
| |||||||||
|
|
context addressable wonk sauce. is the universe quantized? is time discrete or continuous? i suspect the answer is yes |
||||||||
|
|
|||||||||
| |||||||||
|
|
isnt the planck length something to do with it or the fine structure constant or am i talking gibberish |
||||||||
|
|
|||||||||
|
Messageboard index
|
|||||||||
