Fastest Mapping in Soft Samplers

[Burzukh_Studios]

<The following regards drum samples and WAV samples but could be applied to other use cases>

Right now , the fastest sample mapping functionality in a soft sampler that I know and use extensively has to be tx16wx. The ease of use in auto-sorting by name for velocity layers and defining wav matrices for velocity , round robin etc on x and y axis is pretty intuitive and fast.

Calle recently made a slighlty bigger GUI which is a plus too (though it could be bigger yet).

The second one which many of you will laugh at is Reasamplomatic 5000. Yes , this simple thing is actually pretty fast for mapping out drum samples . It allows sorting by file name, RMS or peak value for velocity layers. Multiple instances allow for round robins. I have set up track templates for 2, 3, 4, 5 and 6 round robins which can be accessed with a click.

Both of the above mentioned have no limit (at least theoretically) on the number of velocity layers or round robins. Both work perfectly fine in 64 bit with drag and drop functionality.

What I'm looking for is the next logical level in sampler evolution : intelligent , AI based auto-mapping where I give it a root folder (with child folders) and it asks me some questions / clarifications about the naming convention / structure. Then it loads up the samples into a ready to play template (which requires as little manual tweaking as possible).

Mdrummer is the only software I know that has attempted this ina reasonable manner so far, but it still has a looong way to go in the analysis and automapping of the samples.

I know not of this kontakt that thee speaketh of. Does it have such a functionality?

Can any of the kind and expereinced people here suggest any soft sampler that has a similar functionality ?
I know of only three samplers that I have not treied and all seem massive behemoths with tons of functionality but review are often silent about this bit : Halion, Mach5 and Kontakt .

[Burzukh_Studios]

Any pointers would be highly appreciated. None of the Halion, Mach5 , Kontakt or even NNXT reviews mention anything about intelligent analysis and auto-mapping.

[EvilDragon]

Reviews rarely mention everything, but manuals usually do Both MF3 and Kontakt do have automapping options. MF3 gives you options for automapping when you hold Alt while dropping a bunch of samples in. Kontakt's is also pretty darn good since it uses a token system when you go to Auto Map Setup in Mapping Editor (which I kinda nagged Calle into making for TX16Wx ).

[chickeneps]

Constructor from Chicken Systems has something called the Chizard which is a multi-level "text attracter". You tell it to look for a particular text pattern for any range - key, vel, ksw, ctl sw, rr, or other, and it creates a destination file (any format) for you.

http://www.chickensys.com/support/softw ... izard.html

The elephant in the room, though, is that if you even need to automap stuff, that should be considered in the beginning before the samples even exist. Plan the naming scheme FIRST. That's the neat thing about the Chizard; you can develop the scheme that works best for you (and this includes names of parent folders) and then the Chizard can map it for you without any susceptible pitch or amplitude analysis.

The Chizard has helped quite a few developers do their mapping post-recording.

[Burzukh_Studios]

Reviews rarely mention everything, but manuals usually do Both MF3 and Kontakt do have automapping options. MF3 gives you options for automapping when you hold Alt while dropping a bunch of samples in. Kontakt's is also pretty darn good since it uses a token system when you go to Auto Map Setup in Mapping Editor (which I kinda nagged Calle into making for TX16Wx ).

I was hoping you would chime in regarding this.

Its great to hear that MF3 and Kontakt have auto-mapping.

Is it based on wav file names, where you can define what each part fo the name means and then it does the velocity layering / roundrobin / key mapping ?

[Burzukh_Studios]

Constructor from Chicken Systems has something called the Chizard which is a multi-level "text attracter". You tell it to look for a particular text pattern for any range - key, vel, ksw, ctl sw, rr, or other, and it creates a destination file (any format) for you.

http://www.chickensys.com/support/softw ... izard.html

The elephant in the room, though, is that if you even need to automap stuff, that should be considered in the beginning before the samples even exist. Plan the naming scheme FIRST. That's the neat thing about the Chizard; you can develop the scheme that works best for you (and this includes names of parent folders) and then the Chizard can map it for you without any susceptible pitch or amplitude analysis.

The Chizard has helped quite a few developers do their mapping post-recording.

Thank you for that pointer. I had no clue about this feature, though I have looked at Constructor in the past.

I am talking about when the sampling is not done by the person mapping them, but a clear naming convention exists neverthless. For eg when you buy samples from Impact Soundworks, Drum Drops or Analaogue Drums.

[EvilDragon]

Is it based on wav file names, where you can define what each part fo the name means and then it does the velocity layering / roundrobin / key mapping ?

Yes, Kontakt is a little bit more flexible (same token scheme as in TX16Wx), MF3 only has several naming schemes to apply onto your named sample pool.

The trade-off between them is that in Kontakt you first have to drag your samples in, then run automapping to sort them out (a bit dumb, since you can only drag in 127 samples at a single time...), whereas in MF3 you can do it before the samples are dragged in, by holding Alt when dragging them in, as mentioned in my previous post.

[elcallio]

Since this topic obviously needs to be more about tx16wx, I'd like to ask what sort of mapping features would be most useful? It would be almost trivial to extend the token mapper to operate not on loaded samples, but on stuff gathered from collected paths, and add more token options (create group/program/matrix, assign axis and what not). Building the ui is more painful, but this is how I spontaneously think of it:

- top menu, auto map import
- pick root folder
- either pick stored scheme, or create new
- latter case, pick a sample from which to assign tokens
- assign ops to path parts
- (save preset)
- apply to generate program (s) and sub items

All more or less wizard style.
Would that be a useful tool?

Cheers
/C

[Burzukh_Studios]

Calle the pro-active developer yaaay !

Presently, if I want to get from zero to playable instrument in the shortest time possible, I always used tx16wx, because it is the fastest no-nonsense sampler i know.

My comments are related to drum samples (which is obviously some sort of mental disorder / obsession with me) but maybe can be applied more generally:

- If the token could be mapped to an actual MIDI map that would be very helpful and quick.
for eg : I have groups of multi-velocity samples named stuff like SNR_CENTRE_V1 upto SNR_CENTRE_V20.
Then I would be able to define the token SNR_CENTRE as mapped to MIDI Note Number 38. Auto-mapping would then map all samples conatining the token SNR_CENTRE to MIDI Note 38.

The velocity mapping should also be possible based on "velocity range based on file name sorting" which is already working great when we drag n drop with the correct settings toggled. But such an option is not available currently in the auto-remap token system (which right now only gives the option for lo velocity or hi velocity).

So essentially we drag and drop a bunch of samples into tx16wx. It makes a group which I then auto-remap where i define part fo the name text as an token indicating as to which MIDI key it should be mapped to. The other token tells it to sort into equal velocity ranges sorted by that part of the filename text token.

Folder based would be more interesting

For eg :

I have folder and child folders of such a structure

SAMPLES
------------->KICKS
--------->FELT BEATER
--------->PLASTIC BEATER
-------------->SNARE
--------->CENTER
--------->EDGE
--------->RIM
-------------->RIDE
--------->CENTER
--------->EDGE
--------->BELL
etc
(Note: It could and usually would have waaaay more sub-folders or sub-levels)

In this case , I point the sampler at the first folder (SAMPLES) and it analzyes the full folder structure / child folders.

Then we can assign a MIDI note to each last level child folder in the tree structure (last level in the tree) as well as setting velcoity mapping option for each of these last level child folders (hi / lo or sort into velocity range by filename). Apply and it generates a program with all the mapping ready !

The intelleigent folder based auto-mapping system would make TX16WX the most intelligent soft sampler on earth (not kidding!).

[EvilDragon]

Since this topic obviously needs to be more about tx16wx, I'd like to ask what sort of mapping features would be most useful? It would be almost trivial to extend the token mapper to operate not on loaded samples, but on stuff gathered from collected paths, and add more token options (create group/program/matrix, assign axis and what not). Building the ui is more painful, but this is how I spontaneously think of it:

- top menu, auto map import
- pick root folder
- either pick stored scheme, or create new
- latter case, pick a sample from which to assign tokens
- assign ops to path parts
- (save preset)
- apply to generate program (s) and sub items

All more or less wizard style.
Would that be a useful tool?

Cheers
/C

Excellent ideas there, Calle! I would love to be able to call up this automapping dialog when Alt+dragging a bunch ("bunch" being any number of samples, or FOLDERS containing samples), like in MachFive3. But with token system as it already is. And if you add those options for creating new groups and matrices and stuff, all the better! I would say that with all that it could definitely be the best and fastest automapping built natively into a softsampler ever.

[Burzukh_Studios]

Please see the small attachment (zoom to see the details).
It is just a basic sketch to better illustrate what I'm talking about.

Let me know your thoughts Calle and EvilDragon !

IDEA_SKETCH.jpg

[chickeneps]

Thank you for that pointer. I had no clue about this feature, though I have looked at Constructor in the past. I am talking about when the sampling is not done by the person mapping them, but a clear naming convention exists neverthless. For eg when you buy samples from Impact Soundworks, Drum Drops or Analaogue Drums.

RIght, that's the strength of the Chizard. The auto-mapping ideas presented here are pretty much exactly what the Chizard does in Constructor. You look for a string of text ("token") perhaps in a particular place in the file name (OR parent folder name), and you map it to a key range, velocity range, Keyswitch, Controller Switch (even bend and pressure for esoteric purposes), Round Robin. Of course it's easier when you start from scratch and develop naming conventions of your own, but like you said, in cases where it's given to you by someone else, usually it has logic to it so you can pattern that logic in Chizard and get the things mapped.

[Burzukh_Studios]

Thank you for that pointer. I had no clue about this feature, though I have looked at Constructor in the past. I am talking about when the sampling is not done by the person mapping them, but a clear naming convention exists neverthless. For eg when you buy samples from Impact Soundworks, Drum Drops or Analaogue Drums.

RIght, that's the strength of the Chizard. The auto-mapping ideas presented here are pretty much exactly what the Chizard does in Constructor. You look for a string of text ("token") perhaps in a particular place in the file name (OR parent folder name), and you map it to a key range, velocity range, Keyswitch, Controller Switch (even bend and pressure for esoteric purposes), Round Robin. Of course it's easier when you start from scratch and develop naming conventions of your own, but like you said, in cases where it's given to you by someone else, usually it has logic to it so you can pattern that logic in Chizard and get the things mapped.

It sounds yummy..but i gotta be honest with you : From the feedback around the forums it doesnt seem that constructor works too well, or is much updated /supported.

[chickeneps]

Not true. Why trust secondhand information when you can talk to the source? We fully support it, handle any questions/issues that occur, and it works great. We sell quite a few copies. You can even call us and talk person-to-person 320-235-9798. You can download the demo, you have full access to it.

[Burzukh_Studios]

Not true. Why trust secondhand information when you can talk to the source? We fully support it, handle any questions/issues that occur, and it works great. We sell quite a few copies. You can even call us and talk person-to-person 320-235-9798. You can download the demo, you have full access to it.

Im sorry for putting it bluntly.
Maybe i should try the demo. What is meant by "limited functionality"? Does the "chizard" work ?
How will i test it if i cannot save the format and open it in another sampler ?