KVR Audio

radd · Post by **radd** » Wed Nov 13, 2013 9:07 am

I am using beta9 on a x86 machine with Live8.
Why is the timing so inconsistent?

I've created a quarantined 4/4 loop and the sound comes sometimes way after the MiDi hits, sometimes (strangely enough) before the note will strike. I've bounced the MiDi to audio and you can clearly see how all the transients are messed up.
I was using an Init Preset and there are no LFOs and Attacks are at 0

I'm not using ASIO4ALL

payne_bc · Post by **payne_bc** » Wed Nov 13, 2013 9:28 am

radd wrote:I am using beta9 on a x86 machine with Live8.
Why is the timing so inconsistent?

I've created a quarantined 4/4 loop and the sound comes sometimes way after the MiDi hits, sometimes (strangely enough) before the note will strike. I've bounced the MiDi to audio and you can clearly see how all the transients are messed up.
I was using an Init Preset and there are no LFOs and Attacks are at 0

I'm not using ASIO4ALL

Try the beta6 and see if the timing issue is still there. Then try if it reappears when going to beta7. If that's the case than the package splitting might have something to do with it. Although I work with internal package sizes of 8-9 ms at 44.1/48khz so the difference in timing can never be very big.

H-man · Post by **H-man** » Wed Nov 13, 2013 9:51 am

payne_bc wrote:Still looking...

EDIT: found him! fixed him!

Yep, all dials working with the mouse as expected now

Nice one. Thanks again.

H-man · Post by **H-man** » Wed Nov 13, 2013 10:02 am

chilledpanda wrote:A quick question regarding the noise generator.
The signal flow behaviour of it seems to be if the spline osc. volume is zero then output noise (with matrix mods.) at full whack else mix it with the spline osc. or at least use it's volume is this intentional?

Agree that this could be a bug/oversight. This behaviour is different from that of the Additive Synth which does not incur this "mixing" or "Ducking" effect.

I noticed that this mixing/ducking still occurs if you route noise to ADSR2 in the mod matrix.

IMHO they should be independent for maximum control over each?

payne_bc · Post by **payne_bc** » Wed Nov 13, 2013 10:05 am

H-man wrote:
chilledpanda wrote:A quick question regarding the noise generator.
The signal flow behaviour of it seems to be if the spline osc. volume is zero then output noise (with matrix mods.) at full whack else mix it with the spline osc. or at least use it's volume is this intentional?
Agree that this could be a bug/oversight. This behaviour is different from that of the Additive Synth which does not incur this "mixing" or "Ducking" effect.

I noticed that this mixing/ducking still occurs if you route noise to ADSR2 in the mod matrix.

IMHO they should be independent for maximum control over each?

Hm... I have to try that out. If that is true however I guess I cannot fix it. Since people have probably created patches by now, I might wreck many of those if I change mixing behaviour

H-man · Post by **H-man** » Wed Nov 13, 2013 10:35 am

payne_bc wrote:
H-man wrote:
chilledpanda wrote:A quick question regarding the noise generator.
The signal flow behaviour of it seems to be if the spline osc. volume is zero then output noise (with matrix mods.) at full whack else mix it with the spline osc. or at least use it's volume is this intentional?
Agree that this could be a bug/oversight. This behaviour is different from that of the Additive Synth which does not incur this "mixing" or "Ducking" effect.

I noticed that this mixing/ducking still occurs if you route noise to ADSR2 in the mod matrix.

IMHO they should be independent for maximum control over each?
Hm... I have to try that out. If that is true however I guess I cannot fix it. Since people have probably created patches by now, I might wreck many of those if I change mixing behaviour

Possibly, however because the Spline OSC effectively "kicks out" the Noise (the moment that the Spline OSC volume is raised above zero) the the only way you could hear noise in an existing patch/preset is if you are automating the volume of the Spline OSC on and off, along with a value above zero for the Additive synth

The most likely effect would be noise appearing on patches where previously it was not there...

Anyway, It's your baby so ...just sayin

(PS: I'd be happy to test a beta and copy in the patches that I have so far)

payne_bc · Post by **payne_bc** » Wed Nov 13, 2013 11:37 am

H-man wrote:
payne_bc wrote:
H-man wrote:
chilledpanda wrote:A quick question regarding the noise generator.
The signal flow behaviour of it seems to be if the spline osc. volume is zero then output noise (with matrix mods.) at full whack else mix it with the spline osc. or at least use it's volume is this intentional?
Agree that this could be a bug/oversight. This behaviour is different from that of the Additive Synth which does not incur this "mixing" or "Ducking" effect.

I noticed that this mixing/ducking still occurs if you route noise to ADSR2 in the mod matrix.

IMHO they should be independent for maximum control over each?
Hm... I have to try that out. If that is true however I guess I cannot fix it. Since people have probably created patches by now, I might wreck many of those if I change mixing behaviour
Possibly, however because the Spline OSC effectively "kicks out" the Noise (the moment that the Spline OSC volume is raised above zero) the the only way you could hear noise in an existing patch/preset is if you are automating the volume of the Spline OSC on and off, along with a value above zero for the Additive synth

The most likely effect would be noise appearing on patches where previously it was not there...

Anyway, It's your baby so ...just sayin

(PS: I'd be happy to test a beta and copy in the patches that I have so far)

Yep, it's true. The formula is basically
final = (osc + noise) * osc_volume;
for the cases where oscillator is active, otherwise
final = noise

additive signal is then handled correctly. It's not intentional but this bug came in when I did some optimization quite a while back. It's probably also in the RC from the website. I'll fix that and see how big the impact is on instruments. I suppose the work involved for fixing patches is just adjusting noise volume a bit, that's it.

H-man · Post by **H-man** » Wed Nov 13, 2013 11:45 am

radd wrote:I am using beta9 on a x86 machine with Live8.
Why is the timing so inconsistent?

I've created a quarantined 4/4 loop and the sound comes sometimes way after the MiDi hits, sometimes (strangely enough) before the note will strike. I've bounced the MiDi to audio and you can clearly see how all the transients are messed up.
I was using an Init Preset and there are no LFOs and Attacks are at 0

I'm not using ASIO4ALL

I tried this with the BASSDRUM Patch for 32 bars and every hit is bang on. This preset has ADSR mapped to Volume in the Mod matrix.

Is it possible that this drift in the timing could result from the Volume's default patching, and disappears once the OSC output is routed to the Envelopes?

payne_bc · Post by **payne_bc** » Wed Nov 13, 2013 11:49 am

H-man wrote:
radd wrote:I am using beta9 on a x86 machine with Live8.
Why is the timing so inconsistent?

I've created a quarantined 4/4 loop and the sound comes sometimes way after the MiDi hits, sometimes (strangely enough) before the note will strike. I've bounced the MiDi to audio and you can clearly see how all the transients are messed up.
I was using an Init Preset and there are no LFOs and Attacks are at 0

I'm not using ASIO4ALL
I tried this with the BASSDRUM Patch for 32 bars and every hit is bang on. This preset has ADSR mapped to Volume in the Mod matrix.

Is it possible that this drift in the timing could result from the Volume's default patching, and disappears once the OSC output is routed to the Envelopes?

Hm, I find that unlikely. But it would be good to know by how much the timings are off? Are we talking 10ms? 50ms? 100ms?

chilledpanda · Post by **chilledpanda** » Wed Nov 13, 2013 6:34 pm

payne_bc wrote:
H-man wrote:
payne_bc wrote:
H-man wrote:
chilledpanda wrote:A quick question regarding the noise generator.
The signal flow behaviour of it seems to be if the spline osc. volume is zero then output noise (with matrix mods.) at full whack else mix it with the spline osc. or at least use it's volume is this intentional?
Agree that this could be a bug/oversight. This behaviour is different from that of the Additive Synth which does not incur this "mixing" or "Ducking" effect.

I noticed that this mixing/ducking still occurs if you route noise to ADSR2 in the mod matrix.

IMHO they should be independent for maximum control over each?
Hm... I have to try that out. If that is true however I guess I cannot fix it. Since people have probably created patches by now, I might wreck many of those if I change mixing behaviour
Possibly, however because the Spline OSC effectively "kicks out" the Noise (the moment that the Spline OSC volume is raised above zero) the the only way you could hear noise in an existing patch/preset is if you are automating the volume of the Spline OSC on and off, along with a value above zero for the Additive synth

The most likely effect would be noise appearing on patches where previously it was not there...

Anyway, It's your baby so ...just sayin

(PS: I'd be happy to test a beta and copy in the patches that I have so far)
Yep, it's true. The formula is basically
final = (osc + noise) * osc_volume;
for the cases where oscillator is active, otherwise
final = noise

additive signal is then handled correctly. It's not intentional but this bug came in when I did some optimization quite a while back. It's probably also in the RC from the website. I'll fix that and see how big the impact is on instruments. I suppose the work involved for fixing patches is just adjusting noise volume a bit, that's it.

+1 from me to change it to work separate for more control, currently once you invoke the volume spline osc the noise kinda gets lost anyway so you cannot balance/refine it. I would expect any work for existing patches using the spline/ noise osc. to be minimal rework. I'd be happy with a #ifdef'd bespoke x64 version or a test/beta beta version lol, if you don't want to make the change permanent.

payne_bc · Post by **payne_bc** » Wed Nov 13, 2013 8:22 pm

chilledpanda wrote: +1 from me to change it to work separate for more control, currently once you invoke the volume spline osc the noise kinda gets lost anyway so you cannot balance/refine it. I would expect any work for existing patches using the spline/ noise osc. to be minimal rework. I'd be happy with a #ifdef'd bespoke x64 version or a test/beta beta version lol, if you don't want to make the change permanent.

Well what do you know.. now that I fixed that, the percussion presets actually sound right again

So the bug is not too old.

http://www.tunefish-synth.com/beta11/vst24_tf3_x64.zip
http://www.tunefish-synth.com/beta11/vst24_tf3_x86.zip

payne_bc · Post by **payne_bc** » Wed Nov 13, 2013 8:25 pm

H-man wrote: Sure, it was this one:

1. Map ADSR1 to Noise (or volume)
2. UP the mod value a little
3. Make some quick changes to the sustain on ADSR1
4. Sound drops out, instrument seems to hang.

Off to test Beta10

Hm, this seems fine here with beta11

chilledpanda · Post by **chilledpanda** » Wed Nov 13, 2013 9:39 pm

payne_bc wrote:
chilledpanda wrote: +1 from me to change it to work separate for more control, currently once you invoke the volume spline osc the noise kinda gets lost anyway so you cannot balance/refine it. I would expect any work for existing patches using the spline/ noise osc. to be minimal rework. I'd be happy with a #ifdef'd bespoke x64 version or a test/beta beta version lol, if you don't want to make the change permanent.
Well what do you know.. now that I fixed that, the percussion presets actually sound right again So the bug is not too old.

http://www.tunefish-synth.com/beta11/vst24_tf3_x64.zip
http://www.tunefish-synth.com/beta11/vst24_tf3_x86.zip

Good to hear (party).

Okay I seems I can get a click with the bassdrum (it is pre 11 as well as 11).
If I hold my midi key board for the full length of the bassdrum sound no click. if I give a quick single tap on my keyboard/pad, I get the old click on release (or something related).

payne_bc · Post by **payne_bc** » Wed Nov 13, 2013 9:41 pm

chilledpanda wrote: Okay I seems I can get a click with the bassdrum (it is pre 11 as well as 11).
If I hold my midi key board for the full length of the bassdrum sound no click. if I give a quick single tap on my keyboard/pad, I get the old click on release (or something related).

Yep that ones been reported. I just didn't find it yet

It's caused by the highpass afair

chilledpanda · Post by **chilledpanda** » Wed Nov 13, 2013 10:15 pm

payne_bc wrote:
chilledpanda wrote: Okay I seems I can get a click with the bassdrum (it is pre 11 as well as 11).
If I hold my midi key board for the full length of the bassdrum sound no click. if I give a quick single tap on my keyboard/pad, I get the old click on release (or something related).
Yep that ones been reported. I just didn't find it yet
It's caused by the highpass afair

Ah I wasn't sure. But yes, if I turn HP off on both sounds the click goes.

Did a quick render and loaded the output wav and looks like where the click is the waveform seems to be suddenly inverted?

see image here http://stash.reaper.fm/18711/click.png

I rendered one without high pass, bit harder to tell (more noise), but looks like it doesn't seem to be inverted, if that makes any sense (but I'm possibly talking ollocks

).

EDIT: is actually more obvious when more zoomed out, you can see the sudden inverse ramp (for want of a better description )

One-Synth-Challenge 57: Tunefish - Chat & Gossip