The patch I sent contains a bug. The fir_length is computed wrong (and not clipped to the maximum size allowed by the sinc-tables). This leads to a spurious segfault (on my system this happens nearly always if autosave is triggered).
I will fix this and allow for an even greater oversampling-range (which for the higher factors of course is not usable in realtime-application of lmms but will be for rendering out things...) sorry for the mess, Stefan PS: if I can, I will also optimize this a little bit (it mainly has the status of a working(? erm...) "proof-of-concept". Next patch will be without the mysterious filter-stuff, too btw, as this is not ready anyways... PPS: Toby, I really would appreciate if you could find some time to comment on this. From my point of view this approach makes more sense then any of the previous attempts in reducing the aliasing inside lmms (including the oversampling already available via rendering-out), as it inherently avoids numerical problems with extraordinary high sampling-rates in the (ladspa) plugins as these usually do not assume one could be so "insane" of driving them with multi-megaherz samplerates... PPPS: If you experienced spurious crashes and want a quick-fix: limit the variable "fir_length" in Oscillator.cpp to 128 on both places... ------------------------------------------------------------------------------ Start uncovering the many advantages of virtual appliances and start using them to simplify application deployment and accelerate your shift to cloud computing. http://p.sf.net/sfu/novell-sfdev2dev _______________________________________________ LMMS-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/lmms-devel
