Yofiel all-in-one gen~synth

Wow that is so cool. There is obviously so much attention to detail here; it is very impressive. I will keep an eye on this thread, and will let you know if I find a fix what is causing the midi issue. Could definitely be my own issue. Thanks again!

Well thank you for the feedback. I am running a stress test (see picture) and so far after several thousand notes it is not hanging. However I found a compatibility problem with midi messages and the kslider object. The kslider object issues notes with velocity 0 when the bottom of the key is pressed, and 127 when the top of the key is pressed, which is contrary to normal piano action and a vestige of the graphics object used to make it, which counts pixels from the bottom left corner, so the velocity is backwards. So what I do is subtract velocities from the kslider object from 127, to get a reasonable velocity value. But when playing the kslider object with midi messages, the velocity value is issued uninverted. So the reason you may have seen some notes not playing is that if they have a velocity of 127, I was subtracting them from 127, producing a velocity of zero and hence no sound. So I am adding a bypass for the midi notes, so they don't get subtracted from 127 and play properly.

MIDI protocol supports 31khz, typically with 30 bits for a data message, so messages can arrive within 1ms of each other, which is as fast as the Max scheduler can run. So I checked the scheduler queue, but there doesnt seem to be a problem there, because almost all of the design in in gen~. The most compute-intensive message is a program change, which sets all the dials on the screen, and its possible people with slower machines may have hangs. So I will dig out my old mac mini and do some testing on that too.

Ok, on a 2.5Ghz Mac Mini, I had no midi problems or hangs. The attached screenshot shows synthcore 1.3, which has 3 oscillators instead of two, 15 modulation nodes instead of 12, and 4 new oscillator types. It also shows the test app firing controller changes at maximum midi rate. After severla minutes I stopped the random controller changes and selected a preset, and the synth still worked perfectly.

Activity monitor reported CPU load without the midi stress test was 25%. With the stress test running, cpu load increased 2%. But as expected, the majority of the load was display. With sound running, I switched to the settings tab in the patch, so none of the real-time display of waveforms is being drawn on screen. CPU load drops to under 13%. This compares with my 4Ghz i7 with Nvidia graphics, which reports 5% cpu load with grpahics showing, and 2.5% without the graphics. The cpu profiling in gen~ is much closer to mac activity monitor numbers. On Windows, the reported gen~ load includes graphics rendering overhead, and is reported as a percentage of the throttled cpu clock in power saving mode (which is typically 40% of capacity) and does not take the number of cores into account.

I did find the scope display occasionally stops, which doesn't affect sound, but to remove jitters the scope was set to trigger on signals crossing zero, and in ultra low-frequency and synced oscillators, sometimes the signal does not cross zero a long time. So I added something to fix that, as it may be confusing to users.

I did also find, on the mac, using 'import' messages on buffers no longer works, and also, buffer channels cannot be reliably set via variables in functions unless the wave is imported. So I outcast the functions using variables to set buffer channels where it was a problem, and I declared the buffer data files directly in the buffer object. This is the reverse of how mac handled buffers in quicktime days, when declaring the filename in the buffer object did not always work properly and import messages were more reliable. So it's possible quite a number of people found the wavesets were not making sound, but in absence of more detailed information on when it was not working, I can't be sure.

Mr. wetterberg has confirmed I fixed his 'no filters bug' in private message. I posted it in separate thread, but for those following the original design, the fixed version is here: http://www.yofiel.com/software/mac-and-pc-apps/249-synthcore-2

I am using synthcore20d now, and I have not had any issues. Sounds like maybe that change with kslider fixed it for me. This is awesome, really appreciate the work you have done on this.

Thats great news Brian ) Maybe you will like version 2.3c even more, it has quite a number of new features.

Awesome, I am going to have to check that out. I am loving the presets on there, especially "electric interference." so rad.

Well, I am planning on sharing a few preset banks (they can be loaded from the settings page). So if you want to share any of your creations, feel free to contact me )

I will definitely share anything I happen upon. I am relatively new to this, so not found anything worth sharing yet, hoping to get there though.

I was wondering if there is a way to re-map midi controller settings? I tried editing the code to set the delay 1 feedback to position 6, but that didn't seem to do the trick, as CC6 is still controlling the output mode. Apologize if this is a bad/noob question.

Param
    wheel (0, max=127, min=0)            // 1 mod wheel **
,    d1m (0, max=1, min=0)            // 2 delay 1 mix
,    d2m (0, max=1, min=0)            // 3 delay2 mix
,    foot (0, max=127, min=0)            // 4 foot controller **
,    brange (0, max=127, min=-127)      // 5 bend range **
,    d1z (0, max=0.99, min=0)            // 6 delay 1 feedback


Yes, reordering the params in the massive message at the top of the 'generate lists' subpatch (inside the settings patch) and pressing the indicated button a couple of times will cause the cc# assignments to update automatically. There's alot of comments there indicating how it works.