VSTi -> sf2

[Midiworks]

Hmm maybe you used wrong version ?

I used http://audiolabz.com/vd0321.rar on XP pro r1
Downloaded again, same result.
Did you mabe use a different version ?

If 'detect ...' function is unchecked samples have the same size
http://audiolabz.com/samesize.jpg

Velocity layers i think are created ok, and samples are generated correcly:
http://audiolabz.com/velocity.jpg

And screenshot showing that middle note is sampled for range:
http://audiolabz.com/rootkey.jpg

You pics look all correct in every aspect.

edit:
And if you want select range you must click two times one keyboard to have pink key, the blue one is for single note.
detect sample end causes that samples can have different length because of some small detection errors or different release times.

Thanks for explaning this.
Range works ok now, but still no velo splits...

[CCT]

Hmm did you added notes separately for each added velocity split ?
For each velocity layers different notes can be sampled.

[Midiworks]

Hmm did you added notes separately for each added velocity split ?

No, I set up the velocity ranges, than selected the notes.
(pink)

For each velocity layers different notes can be sampled.

Very unusual...
btw. velocity ranges can not overlap, sad,
this should be possible.

[CCT]

So tell me the best solution that would be usable

edit:
hmm i will add a switch like 'use notes selected for first layer'

and now i am working on VST chaining

Greetz

[Midiworks]

So tell me the best solution that would be usable

Ok, I try.

For each velocity layers different notes can be sampled

This does not make much practical sense to me.
I would drop that or add an on/off option,
but as standard I would disable this.

And if you want select range you must click two times one keyboard to have pink key, the blue one is for single note.

I would stick to 1 color but add a button to:

Multisample
Single note
Chord

Much clearer !

-----------------------------------------------------
Velocity splits like this for the sf2 dump are fine.
But many times you want the "program" to overlap the
velocity splits.
This way the sound does not switch hard from one
wav top the next but changes smooth into the next one.

A velocity off set function would be good for this.
Like:
Velocity splits 0-64
Velocity off set +/-10 (new button)

This means the sample dump still does 0-64
but the layer would play from 0-74

Or sample dump=65-95,
the layer would play velocity 55-105
(0=always 0, 127=127, no offset for those)

I hope you understand....

[duke4e]

great idea for a program!
hehe, i already see all my favorite VSTi burned on cd in sf2 format

could you implement BPM settings (some VSTi's use host bpm to sync with their internal calculations)

[CCT]

Midiworks: i understanded

duke4e: that feature will be in next version - be patient. With 5-7 days i release new version with VSTi & VST chaining and maybe with xfade looping.

Greetz.

[Sascha Franck]

maybe with xfade looping.

If this will work: WOW-f**king-WOW!
Really, that would be a quantum leap forward in sampling VSTis, at least for me!
Having to load each and every sample separately into WaveLab (or whatever editor) is really a nightmare, once you've done more or less complexed patches.

OK, here's a suggestion, should you make that x-fade looping happen:
I assume there will be some loop start/end point setting (otherwise it'd be really problematic to find proper looppoints), to find out the best setting for those, it could be great to have some sort of "sound progress bar". That is, when you press a key, some "counter" would start, displaying seconds and milliseconds. So, on sounds with a strong attack phase you could just watch that counter and note the time when the attack phase was over, just to memorize it in order to find better loop points.
I don't know whether that's possible, but it surely would be great.

[Caleb]

I admire your pluck to try and put crossfading looping into this.

I will certainly be a dedicated customer if you can implement something like that.

I really hate sampling my VSTi and then looping everything painstakingly just so I can load it up into a sample synth and mangle it further.

This could be the answer to all my prayers.

Caleb

[tl]

Looks like a very promising tool.

A question:
How do switch modes (multisample/chord) and select the output format?
Clicking/Dragging the buttons with the mouse doesn't do anything.

tl.

[Midiworks]

With 5-7 days i release new version with VSTi & VST chaining and maybe with xfade looping.

There is a general misunderstanding about
xfade vs crossfade.

These are NOT the same things.
(as the name already indicates. )

xfade is usefull for very short sample loops. (1-2sec.)
Or for blends between audio tracks for a click free playback. (in ms)

crossfade is great for longer samples (3sec and up)

For todays huge multisampled instruments or for VSTi Dumper,
I would drop the idea of xfade completly.

OK, here's a suggestion, should you make that x-fade looping happen:
I assume there will be some loop start/end point setting (otherwise it'd be really problematic to find proper looppoints), to find out the best setting for those, it could be great to have some sort of "sound progress bar". That is, when you press a key, some "counter" would start, displaying seconds and milliseconds. So, on sounds with a strong attack phase you could just watch that counter and note the time when the attack phase was over, just to memorize it in order to find better loop points.

I can not agree at all with Sascha.
The only good solution I have seen is,
DiscoDSP's Highlife.
http://www.discodsp.com/highlife/

The secret to success is to use a fixed crossfade lenght for a fixed sample duration.
Aplied to ALL multisamples equally.
Or calculated dynamically,
from an initial sample lenght.
(then no fixed lenght must be used.)

While a variation of this would be great.
lower samples=longer crossfade,
higher samples=shorter crossfade.


To be continued !
Stay tuned.

[Sascha Franck]

Ahem, Midiworks, apart from agreeing with the rest of your post, xfade is exactly the same as crossfade. Actually xfade is a shortform for cross fade (x = cross).
In fact, both do the same. It's just a matter of the terminology you prefer, not a matter of milliseconds or so.

Anyways, maybe a crossfade setting of 50% would be a good thing to start with, in case it should be automated (of course a user defineable valus should be there too). With 50% it should be that half of the loop is "mirrored" (sorry, found no better expression). That approach usually gives perfect loop points, the only problem being the attack phase until the loop starts.
Also, it might be nice to have a few more options:
- Cut release samples. This means, all samples right from the loop range will be deleted. I often do that because I will rebuild the release in my sampler anyways.
- Cut "attack" samples. This means, all samples left from the loop range will be deleted. Sometimes I only want the "pure" sound, without any attack made up by the sampled VSTi. I will rebuild most of my envelopes later on anyways.
- Trim samples by loops. Which would just combine the two things.

[Midiworks]

Sascha: http://www.dehaupt.com/SynthEdit/SampleSnatcher%20I.htm

This was a lot of work to build this with Dave Haupt.
I mean he did the coding, but to make him understand,
how things are working and should work.
But now it is great.
You can have a fader and by moving it aply the crossfade in realtime !

Live audio grabber and looper

Stereo sample recorder/player can produce smooth, crossfaded loops on the fly from your audio input. Automatically finds zero crossings to minimize clicks. Recording can be set to start and stop automatically, with adjustable sensitivity to input. Playback pitch/speed is continuously variable. Looping can be forward, backward, or alternating. Start, loop, end, and current read positions can be manipulated directly, and even modulated at audio frequencies. Handles samples of up to 3 minutes.

[Sascha Franck]

Sascha: http://www.dehaupt.com/SynthEdit/SampleSnatcher%20I.htm

That's reading just like the right thing!
Unfortunately I don't know anything about SE, so I wouldn't exactly be able to check it out quickly.
Seems as if this would be what is needed for VSTi dump.

Hm, someone should integrate the same principle into Audacity or something else. Maybe I should contact the Audacity dude.

And there should be freaking batch processing of EXACTLY the same thing in WaveLab or so. I do ALL my loop adjustments manually - it's driving me mad!

[Midiworks]

Sascha, I will provide all the help to CCT he might need.

VSTi Dumper has a lot of potential.
I will do my best to help to make it a great and unique program.