![]() |
![]() ![]() ![]() ![]() ![]()
|
next newest topic | next oldest topic |
Author | Topic: Aliasing |
Olle Romo Member |
![]() ![]() ![]() This is probably a real duh question, but I can't seem to be able to find anything on the subject in the manual or here. What would be a good way to deal with the aliasing noise from - say - a saw tooth oscillator in the upper registers? Without extensive multi sampling. Or filtering. Can I algorithmically (if that's a word) generate a gorgeous sawtooth without having to read a wave table? One that's smooth from bottom up? IP: Logged |
SSC Administrator |
![]() ![]() ![]() How about additive synthesis? The amplitude of each harmonic n is 1/n. You can use a SpectrumFromArray Sound to create the spectrum for an Oscillator bank. In the prototypes check out "Saw Square Additive Lite". Prototypes using the multicycle approach include "Oscillator (sawtooth)" and "Oscillator Saw Multicycle KBD". Use Control+B and type "saw" to find these examples. IP: Logged |
Olle Romo Member |
![]() ![]() ![]() Thank you for your suggestions. I've come across quite a few sounds that do the analog wave thing well. I'm wondering if there's a way to generate a wave in smalltalk in realtime and have access to parameters within the algorithm in realtime? Or would it have to happen in assembler? I did download a sheet on Motorola assembler and it's the stuff bad dreams are made of... Max/MSP speak proudly of their saw generator as one that doesn't alias. How is this done in Kyma without wavetables? Is additive the best way? Are there any other methods to generate basic synth waveforms? I've had my system for two years now and I'm finally beginning to penetrate the first layer in the Kyma onion... trying to keep my eyes from watering ![]() IP: Logged |
SSC Administrator |
![]() ![]() ![]() "I'm wondering if there's a way to generate a wave in smalltalk in realtime and have access to parameters within the algorithm in realtime?" In fact, that is what *every* Kyma Sound is doing: generating a wave in realtime with realtime access to the parameters of the algorithm. (In this case, the algorithm is 'written' as a graphical signal flow diagram instead of lines of text.) If you want to store that signal as a static "wave", you can use a MemoryWriter to write the signal into Capybara RAM. Then you could read it from any of the modules that read memory (like Oscillator, Sample, SampleCloud, etc) Another approach to generating a waveform is the XenOscillator; you supply endpoints and the XenOscillator connects the endpoints with stairstep, linear, or spline interpolation. The endpoints can be generated algorithmically using CapyTalk. Pete's Surbiton Oscillator might also be an interesting one to look at. Depending on the frequency of the oscillator, he prefilters the waveform at the half sample rate before writing it into the wavetable. Additive synthesis with the SyntheticSpectrumFrom Array is also a method for 'writing' a waveform in real time while adjusting parameters such as the amplitudes and frequencies of the partials. IP: Logged |
Olle Romo Member |
![]() ![]() ![]() Thank you so much for the replys. I'm trying to start up the surbiton sounds but they all ask for soundfiles like '0.35423' and the like. It looks like that is the name it uses for the wave it writes into RAM but it won't start up without finding it on disk and I don't have it. What do I do? Are there any more immersion weekends planned? IP: Logged |
SSC Administrator |
![]() ![]() ![]() Is there a checkbox labelled "From Memory Writer"? If so, that should be checked in the Oscillator. (That parameter was added sometime after Pete made the Surbiton examples) IP: Logged |
Olle Romo Member |
![]() ![]() ![]() There is. I'm trying the "Square - SurbitonOscillator" in "Analog KBD.kym". After expanding the sound I find two oscillators with the 'fromMemoryWriter" checkbox but I get a bunch of errors when I check them. The XenOscillator rocks! ![]() IP: Logged |
pete Member |
![]() ![]() ![]() Hi Olle The surbiton oscilator in the prototypes is a limited encapsulate saw tooth only version. I latter opened it up and posted it on the forum in it's raw state. http://www.symbolicsound.com/cgi-bin/forumdisplay.cgi?action=displayprivate&number=3&topic=000049 Here you can generate any waveform you like but you must allways make it sample rate /4096 in pitch. The surbiton osc itself then repitches and filters the signal to avoid aliasing. Hope it helps Pete IP: Logged |
Olle Romo Member |
![]() ![]() ![]() Hi Pete, This is what I was looking for!! Smooth from top to bottom, no multi sample boundaries and easy on the processors. Fab! ![]() IP: Logged |
SeanFlannery Member |
![]() ![]() ![]() Hi SSC, Probably one of my biggest wishes for Kyma would be non-aliasing oscliators. Apart from the surbiton and the oscilators containing the decription 'non-aliasing', lots of the other ones present aliasing to some degree - the Xen oscilators, the conventional oscilators, the GAOscilators. I dont know how difficult eradicating aliasing would be - I suspect it's not child's play - but is it able to be reduced utilising the new hardware platform? All the best, IP: Logged |
SeanFlannery Member |
![]() ![]() ![]() Also, while digging around researching aliasing in Kyma I came across a tool Luddy had put together that generates Wavetables suitable for use in the non-aliasing oscilator. Unfortunately this only works on the Intel Mac platform. http://www.symbolicsound.com/cgi-bin/forumdisplay.cgi?action=displayprivate&number=3&topic=000186 It would be soooo cool to be able to use that tool on a PC. Luddy, you mentioned on that post you may develop it further if there was interest, mark me down as definitely interested please Regards, IP: Logged |
All times are CT (US) | next newest topic | next oldest topic |
![]() ![]() |
This forum is provided solely for the support and edification of the customers of Symbolic Sound Corporation.