Session Drummer 3 - The Problem Of Poor Editing/Programming

OK, the first part in my soon to be multi-part rant. Maybe not for the faint of heart. For those reading please do us all one big favour and read very closely before complicating the issue with responses like "well, just draw in a velocity..." because if you follow closely you'll realise that just ain't an option. Unless you're rather familiar with programming full sample sets from scratch you likely aren't going to find this entertaining either. This is targeted only at Session Drummer as it is the included plugin also sold by Cakewalk however this is sadly a problem with WAY to many sample sets.
 
I sincerely hope that some of the sample companies that represent themselves here on occasion will consider chiming in and setting me straight where/when they can.
 
Why does this bug me? This is not rocket science, this is not trying to get core audio latency to work with multiple types of RAM, different video cards, hard drives, motherboards blah blah blah. This is simple multi-layer sample editing. If I have to waste time fixing a computer or solving driver issues because of the choice in soundcards I made, that's one thing. But when I have to spend stupid amounts of time fixing something when I'm in the middle of working, something that should be friggin' simple for the person who created the SFZ files, then that's a problem. And this particular problem happens way, way too often. Who exactly is it programming this crap and why is Cakewalk allowing it considering what an easy fix it should be?
 
 
So, take Session Drummer 3 for example, the Steven Slate Old Zep Kit. Great sounding samples, no doubt. Start programming them in Sonar...ouch, really poor mapping problems are immediately apparant. Let's look only at the snare to simplify.
 
To make this as easy (not likely) to follow as I can let me clearly illustrate a single part of the larger issue. To set up this test I simply created 127 midi snare hits ranging in velocity from 127 down to 0. Properly programmed (or "edited" if you prefer) samples should have their volume in dB lower, as the velocity lowers, in a relatively predicatable way whether it be true linear or in an exponential (?) manner as in a velocity curve.
 
A change of one step in velocity should -not- cause a drop in volume of over 12dB, that's some really poor programming and might I add some pretty poor quality control that this has not been addressed in over a year of use by probably tens of thousands of people. Do people not hear this? Do they not realize what they're not hearing and that they are losing the benefit of these great multi-velocity, multi-layered hits with 4 levels of randomization? YIKES! If I was the sample developer I'd be p****d.
 
Witness Number One please step up:
http://www.flickr.com/photos/81774270@N03/
 
You can click on each picture and on the next page there are zoom options which may be helpful.
 
SSlate1 shows the snare hits descending from a velocity of 127 to 0 using the standard SFZ snare file included with Sonar. One can quite clearly see the poor ramping of the velocity regions as they switch from one level to another. Incidentally, for those interested, the initial hits up to the 21 second mark are actually two velocity regions which -are- set up well. The next three velocity regions are bad and clearly visible.
 
Now SSlate2. Just after 22 seconds it can be seen that a snare with a velocity of 79 would produce a hit about -10.4 dB where a velocity of 78 would produce a hit of about -17dB which is ridiculous. This clearly shows the crossover from one velocity map to another.
 
SSlate3 shows it just keeps getting worse. The cursor is set at around 42 seconds, just before a hit that resulted from a velocity of 26. The previous hit was a velocity of 27, only -one- velocity step away. The result of one velocity step is a bare minimum of 9dB although in fact a little worse, on the left side it is clearly about 13 dB.
 
SSlate Test 1 is an example of the same test after editing the velocity responses in notepad. Here you can see that the velocity stepping between regions is far smoother and almost useable.
 
This "editing", however, turns out to be a bigger nightmare than you can imagine as the info available
http://www.cakewalk.com/DevXchange/article.aspx?aid=108
is (wow, big suprise here...) inaccurate and vague and the example files do not reasonably reflect the protocols. For example, the "Amplifier" section quite clearly details that the volume= command has a range of -144 to +6 dB and yet it was necessary to use values over +100 to "fix" the Slate snare drum, a value that is not even supposed to be valid (yet I clearly proved it was). Great...
 
Anyways, next up, a clear description of -how- and -why- this affects you, the user/consumer/musician (in case you haven't already figured it out). I also intend to discuss whether or not normalization of all samples in a set is a good idea, functional velocity ranges versus functional SPL levels and if there should be some sort of standard (yes) in sample sets whether manufacturers adhere to these recommendations or not, amongst other issues.
 
Edited to change "layers" to "regions" in a few spots.

i think its great you find it neccessary to start a whole new thread on something you just had a thread on and i think its great that i offer a solution for drawing in velocitys in the piano roll and you pretty much mock me by saying in your first paragraph, i complicated the issue by suggesting
"well, just draw in a velocity..."
 
most proffessionals use something like bfd..or addictive drums...or superior drummer.
though,i still find session drummer 3 my vsti kit of chose because of its sounds and flexibility.
in ease of use and velocity.

i find its velocitys change very nicely while using the piano roll with the smart tool.
but then again,im not a scientist probing through its sample with a pencil in a peetree dish. 
i think session drummer is auesome,sure it could use some beefing up and some fixes here and there.

but dude,this thread is a full blown rant.
once i read the part of you calling wittnesses to the stand ,i was like.. "wow.."..lol;
heres how you can tell the difference between a rant and not a rant...when words not allowed on the site
are replaced with things like ****.
thats usualy a tell tale sign. 
 
i believe there are ways to exspress your concerns and then there are ways to sound like
"i am so mad im going to start a thread...no im going to start multiple threads".

take some deep breaths,its only a vsti-
you can always get another one if you dont like it.

i do like it...have you tried the steven slate sizzle kit?

SToons. Are you this hard on real drummers?

i think ive got a great sence of hummor,didnt you read my post?...lol.
due,this is precisely what im talking about,

Let me ask you this - if a hit with a velocity of  79 produces a volume of -10.4 dB and a hit of 78 creates a volume of -17 dB then how can you possibly get a hit with a volume of -15dB? How about -14dB? -12dB?

the above suggest your qoute as that of a scientist.
the mojority of us are musicians.
i bet more then half the forum reads something like this and interpets this as...bla..bla..bla.

you have a small handfull of followers here and even this one guy saying he's experienced it to but then suggests he didnt know the sfz's could be edited manualy.
i have no idea what that ment.

i just think your being a little picky.
if this was such an out cry of a problem,why now?..5 years after sd3 has been out this is now surfacing.
is it just youve been data collecting for the past 5 years and now its time to post your finding?

i respect what your trying to do here.
i also respect the fact youve taken the time to try and bring up this issue to make things better in sd3.
i just think your angry and when your angry things dont come out so well.

but if session drummer is not giving you the results you want,maybe its just not able to do this.maybe it never was ment to be a perfect velocity virtual drum kit.
maybe it is what it is and thats it.
remember,sd3 never advertised the velocitys were perfect.
but its us as users that take for grantite so much,things like this...we expect these things. 

i think more than anything,arent you curious that when changing a hi-hat you lose the open hat?.
there are many things sd3 could improve on.
ahh,maybe i should just bow out now.
i guess its not really fair,this is your thread.
it started out with my opinion and now its taking a new direction.
the last thing i want to do is start attacking your theorys..maybe to late..

in a sence its people like you that make this software better because their complaints open up the eyes of staff which trickles down to programers.
but there is a way of going about it which includes trying to be a gentlemen about it and leaving emotions at the door.

like i said,i will bow out of this one and i hope you find what your looking for in this thread.

good luck.

newbie_hobbyist


Yes, I've noticed this exact same issue, although I hadn't figured out you can edit .sfz's manually.

If anyone has taken the time to fix some .sfz's, would you mind sharing them (hopefully not illegal since they don't contain any actual sample data)?

Its got nothing to do with fixing the .sfz's.   Session Drummer, is a multi-sample rompler/sampler.   Sampling is an art form, in and of itself.  Drum's are of the most difficult instruments to sample, because their natural adsr envelope is extremely complex.

Drum sampling for a product like Session Drummer, BFD, Addictive Drums, is fairly different than lifting a break off a record, slicing it up in Beatscape, and playing it back as a chopped loop.

Drum sampling for the Session Drummer type of product should be, "multi-sampled recreation of hopefully high end drum kits, processed in an extremely cool ways, to give them an excellent sound."

Fixing an .sfz file, will not fix a sampling job where: you simply do not have enough velocity layers.  the sounds themselves are recorded in no special way.  the sounds are not processed well.  the sounds come from a poor source instrument.  Ofcourse, all that falls apart if the .sfz is poorly written, but from what I understand about .sfz is that its a simple type of file to write.  Akin to html, in its simplicity, and certainly not as complex as low level computer language.

Session Drummer is a NO COST add on in Sonar X1 software.  I'm not sure what kind of complaints we should be having.  The software is coded correctly, functions correctly, and with the proper content is going to sound amazing.  That said, "the factory content provided for session drummer could use a total rethink."  And I specifically mean that...  A.  There isn't much there.  B.  The Steven Slate kits are supposed to be excellent, as Steven Slate tries to markets his drums as, "the best around."  B.  The Sonic Reality kits are supposed to be good.

I know there is some stuff from the guys at Cakewalk, and Roland.  I'm not going to be down on that content!  I think its cool that content is in there.  I don't want to shuffle negativity around.

Even with A, B, and C; we are still talking about a Rompler/Sampler/VST that is provided free as charge.   Great, there's tons of stuff out there.  BFD--->Battery and everything in between.

You will still find the need to use professionally sampled aftermarket content.  Technically BFD really doesn't do much more than Session Drummer. Its got a few things..  A.  Its factory content is way better.  Still not great.  On the level of Battery, maybe a little better.  C.  The ability to effect the sound of the kits via some features of its design..  damping, bleed, moving mics, stuff like that.

The one thing that is the same across BFD, Session Drummer, and Addictive Drums and the like is that the content must be excellent to get an excellent result!

When you load the Platinum Samples Andy John Kit into Session Drummer, you see:  A.  Session Drummer works great.  B.  .SFZ works great.  C. Excellent content works great.  And, Platinum is coming with more Session Drummer content!  Hooray cause we will all be saved.

If there was better content for Session Drummer.  Well sampled, with multiple velocity layers.  Not 4.  Not 2.  But 12, 15, or 20..   If there was great content, and a lot of it..  We wouldn't be having these threads.

Right now, if you need a drum solution..  And all you've got is Session Drummer... You are in a rough spot.  If you are brand new, have only Session Drummer, but a little $$  You might want to head over to www.native-instruments.com and demo battery or something.  maybe go to www.fxpansion.com and check out bfd...

battery is less expensive btw!


Cheers

To get a better curve you need more velocity levels...  you don't need 128 but 16 is good.  Since the VI has to velocity scale the samples between the notes defined for each sample if you only have a few samples the curve isn't going to be as good as f you have more samples with smaller sections being scaled.

Our beta testers are very happy with our kits we're working on which have 16 velocity levels and 4 round robins per level.

In the SFZ you can adjust the curve of the scaling... but as common sense would dictate... more is better... the balance is between RAM usage and playability.

Cheers,

Rail
---------------------
www.platinumsamples.com

Platinum Samples


To get a better curve you need more velocity levels...  you don't need 128 but 16 is good.  Since the VI has to velocity scale the samples between the notes defined for each sample if you only have a few samples the curve isn't going to be as good as f you have more samples with smaller sections being scaled.

Our beta testers are very happy without our kits we're working on which have 16 velocity levels and 4 round robins per level.

In the SFZ you can adjust the curve of the scaling... but as common sense would dictate... more is better... the balance is between RAM usage and playability.

Cheers,

Rail
---------------------
www.platinumsamples.com

Thanks for chiming in.
 
Yes, you are correct that more velocity regions would allow more flexibility of the curve. However I clearly demonstrated that with a little tweaking I was able to vastly improve the response of the regions and I have subsequently improved it since. Why? Why should I have to? Why was the original so poorly programmed? Why were glaring descrepencies left this long and how come the programmer didn't hear this when it's so obvious? How can someone not notice when a soft hit after a loud hit causes the loud hit to cut-off like a gated snare due to poor release, polyphony or overlap values (which obviously has nothing to do with how many regions or layers there are). Fact is, a single sample region is a joke to make respond properly to velocities - the more regions there are, the harder to properly program regions.
 
Why are so many sample libraries poorly programmed? Never heard a single dud in Gigasampler/Gigastudio. If you like we can discuss Battery Studio drums (same problem) or any number of other products with the same problems.
 
This is the issue.

//  sfz definition file
//  (c) Zyler Vega Cakewalk, Inc 2009
//  Steven Slate - Old Zep Kit



//========================rimshot
<group>
key=37
group=37
polyphony=1
rt_decay=6.500
ampeg_release=2

<region>
sample=StevenSlate Snares\SideStickZ1Soft1.wav
lovel=0
hivel=43
lorand=0.0
hirand=0.25
<region>
sample=StevenSlate Snares\SideStickZ1Soft2.wav
lovel=0
hivel=43
lorand=0.25
hirand=0.5
<region>
sample=StevenSlate Snares\SideStickZ1Soft3.wav
lovel=0
hivel=43
lorand=0.5
hirand=0.75
<region>
sample=StevenSlate Snares\SideStickZ1Soft4.wav
lovel=0
hivel=43
lorand=0.75
hirand=1.0

//---

<region>
sample=StevenSlate Snares\SideStickZ1Medium1.wav
lovel=44
hivel=86
lorand=0.0
hirand=0.25
<region>
sample=StevenSlate Snares\SideStickZ1Medium2.wav
lovel=44
hivel=86
lorand=0.25
hirand=0.5
<region>
sample=StevenSlate Snares\SideStickZ1Medium3.wav
lovel=44
hivel=86
lorand=0.5
hirand=0.75
<region>
sample=StevenSlate Snares\SideStickZ1Medium4.wav
lovel=44
hivel=86
lorand=0.75
hirand=1.0

//---

<region>
sample=StevenSlate Snares\SideStickZ1Hard1.wav
lovel=87
hivel=127
lorand=0.0
hirand=0.25
<region>
sample=StevenSlate Snares\SideStickZ1Hard2.wav
lovel=87
hivel=127
lorand=0.25
hirand=0.5
<region>
sample=StevenSlate Snares\SideStickZ1Hard3.wav
lovel=87
hivel=127
lorand=0.5
hirand=0.75
<region>
sample=StevenSlate Snares\SideStickZ1Hard4.wav
lovel=87
hivel=127
lorand=0.75
hirand=1.0



//========================snare1
<group>
key=38
group=38
polyphony=1
rt_decay=6.500
ampeg_release=2

<region>
sample=StevenSlate Snares\OldZepSnareZ1UltraSoft1.wav
lovel=0
hivel=26
lorand=0.0
hirand=0.25
<region>
sample=StevenSlate Snares\OldZepSnareZ1UltraSoft2.wav
lovel=0
hivel=26
lorand=0.25
hirand=0.5
<region>
sample=StevenSlate Snares\OldZepSnareZ1UltraSoft3.wav
lovel=0
hivel=26
lorand=0.5
hirand=0.75
<region>
sample=StevenSlate Snares\OldZepSnareZ1UltraSoft4.wav
lovel=0
hivel=26
lorand=0.75
hirand=1.0

//---

<region>
sample=StevenSlate Snares\OldZepSnareZ1Soft1.wav
lovel=27
hivel=51
lorand=0.0
hirand=0.25
<region>
sample=StevenSlate Snares\OldZepSnareZ1Soft2.wav
lovel=27
hivel=51
lorand=0.25
hirand=0.5
<region>
sample=StevenSlate Snares\OldZepSnareZ1Soft3.wav
lovel=27
hivel=51
lorand=0.5
hirand=0.75
<region>
sample=StevenSlate Snares\OldZepSnareZ1Soft4.wav
lovel=27
hivel=51
lorand=0.75
hirand=1.0

//---

<region>
sample=StevenSlate Snares\OldZepSnareZ1Medium1.wav
lovel=52
hivel=77
lorand=0.0
hirand=0.25
<region>
sample=StevenSlate Snares\OldZepSnareZ1Medium2.wav
lovel=52
hivel=77
lorand=0.25
hirand=0.5
<region>
sample=StevenSlate Snares\OldZepSnareZ1Medium3.wav
lovel=52
hivel=77
lorand=0.5
hirand=0.75
<region>
sample=StevenSlate Snares\OldZepSnareZ1Medium4.wav
lovel=52
hivel=77
lorand=0.75
hirand=1.0

//---

<region>
sample=StevenSlate Snares\OldZepSnareZ1Hard1.wav
lovel=78
hivel=102
lorand=0.0
hirand=0.25
<region>
sample=StevenSlate Snares\OldZepSnareZ1Hard2.wav
lovel=78
hivel=102
lorand=0.25
hirand=0.5
<region>
sample=StevenSlate Snares\OldZepSnareZ1Hard3.wav
lovel=78
hivel=102
lorand=0.5
hirand=0.75
<region>
sample=StevenSlate Snares\OldZepSnareZ1Hard4.wav
lovel=78
hivel=102
lorand=0.75
hirand=1.0

//---

<region>
sample=StevenSlate Snares\OldZepSnareZ1Crack1.wav
lovel=103
hivel=127
lorand=0.0
hirand=0.25
<region>
sample=StevenSlate Snares\OldZepSnareZ1Crack2.wav
lovel=103
hivel=127
lorand=0.25
hirand=0.5
<region>
sample=StevenSlate Snares\OldZepSnareZ1Crack3.wav
lovel=103
hivel=127
lorand=0.5
hirand=0.75
<region>
sample=StevenSlate Snares\OldZepSnareZ1Crack4.wav
lovel=103
hivel=127
lorand=0.75
hirand=1.0



//========================snare2
<group>
key=40
group=38
polyphony=1
rt_decay=6.500
ampeg_release=2

<region>
sample=StevenSlate Snares\OldZepSnareZ1UltraSoft1.wav
lovel=0
hivel=26
lorand=0.0
hirand=0.25
<region>
sample=StevenSlate Snares\OldZepSnareZ1UltraSoft2.wav
lovel=0
hivel=26
lorand=0.25
hirand=0.5
<region>
sample=StevenSlate Snares\OldZepSnareZ1UltraSoft3.wav
lovel=0
hivel=26
lorand=0.5
hirand=0.75
<region>
sample=StevenSlate Snares\OldZepSnareZ1UltraSoft4.wav
lovel=0
hivel=26
lorand=0.75
hirand=1.0

//---

<region>
sample=StevenSlate Snares\OldZepSnareZ1Soft1.wav
lovel=27
hivel=51
lorand=0.0
hirand=0.25
<region>
sample=StevenSlate Snares\OldZepSnareZ1Soft2.wav
lovel=27
hivel=51
lorand=0.25
hirand=0.5
<region>
sample=StevenSlate Snares\OldZepSnareZ1Soft3.wav
lovel=27
hivel=51
lorand=0.5
hirand=0.75
<region>
sample=StevenSlate Snares\OldZepSnareZ1Soft4.wav
lovel=27
hivel=51
lorand=0.75
hirand=1.0

//---

<region>
sample=StevenSlate Snares\OldZepSnareZ1Medium1.wav
lovel=52
hivel=77
lorand=0.0
hirand=0.25
<region>
sample=StevenSlate Snares\OldZepSnareZ1Medium2.wav
lovel=52
hivel=77
lorand=0.25
hirand=0.5
<region>
sample=StevenSlate Snares\OldZepSnareZ1Medium3.wav
lovel=52
hivel=77
lorand=0.5
hirand=0.75
<region>
sample=StevenSlate Snares\OldZepSnareZ1Medium4.wav
lovel=52
hivel=77
lorand=0.75
hirand=1.0

//---

<region>
sample=StevenSlate Snares\OldZepSnareZ1Hard1.wav
lovel=78
hivel=102
lorand=0.0
hirand=0.25
<region>
sample=StevenSlate Snares\OldZepSnareZ1Hard2.wav
lovel=78
hivel=102
lorand=0.25
hirand=0.5
<region>
sample=StevenSlate Snares\OldZepSnareZ1Hard3.wav
lovel=78
hivel=102
lorand=0.5
hirand=0.75
<region>
sample=StevenSlate Snares\OldZepSnareZ1Hard4.wav
lovel=78
hivel=102
lorand=0.75
hirand=1.0

//---

<region>
sample=StevenSlate Snares\OldZepSnareZ1Crack1.wav
lovel=103
hivel=127
lorand=0.0
hirand=0.25
<region>
sample=StevenSlate Snares\OldZepSnareZ1Crack2.wav
lovel=103
hivel=127
lorand=0.25
hirand=0.5
<region>
sample=StevenSlate Snares\OldZepSnareZ1Crack3.wav
lovel=103
hivel=127
lorand=0.5
hirand=0.75
<region>
sample=StevenSlate Snares\OldZepSnareZ1Crack4.wav
lovel=103
hivel=127
lorand=0.75
hirand=1.0



//========================snareroll
<group>
key=39
group=38
off_by=38
rt_decay=6.500
ampeg_release=2
amp_veltrack=100


<region>
sample=StevenSlate Snares\OldZepSnareZ1zRolls1.wav
lovel=0
hivel=127
lorand=0.0
hirand=0.25
<region>
sample=StevenSlate Snares\OldZepSnareZ1zRolls2.wav
lovel=0
hivel=127
lorand=0.25
hirand=0.5
<region>
sample=StevenSlate Snares\OldZepSnareZ1zRolls3.wav
lovel=0
hivel=127
lorand=0.5
hirand=0.75
<region>
sample=StevenSlate Snares\OldZepSnareZ1zRolls4.wav
lovel=0
hivel=127
lorand=0.75
hirand=1.0

Woah, dude, that was FAST!!! Thanks! Well, after scanning for a whole 2 seconds the answer is no, it hasn't ben fixed.
 
I will post a very obvious audio example later of why this is an issue.
 
As for the drummer comment, I, uhm, err, ahh, unhhhhhh....how many drummers does it take to screw in a lightbulb?
 
<ST reverts to diversionary tactics...>

SToons

 I have a notoriously thick skull. 

+ 1...

:-)


Post the altered .sfz file.  Let the users load it up in their own copy of Session Drummer.  An audio example, has little meaning... it does not  allow us to test the new sfz, and see if it really makes for a better velocity curve.

I have my doubts.  I believe the problem is the Steven Slate kits are pretty bad imo.  Having, "one excellent snare drum," does not a good sampled drum kit make!


Cheers

Actually, the only hard thing about sampling drums is that you need tons of mics to record the direct (close mics), OHs and room mic samples keeping them all in phase..... once set up though, its just hit the record button.

Jim

Linear Phase

SToons

Linear Phase

SToons

I have a notoriously thick skull. 

+ 1...

:-)


Post the altered .sfz file.  Let the users load it up in their own copy of Session Drummer.  An audio example, has little meaning... it does not  allow us to test the new sfz, and see if it really makes for a better velocity curve.

I have my doubts.  I believe the problem is the Steven Slate kits are pretty bad imo.  Having, "one excellent snare drum," does not a good sampled drum kit make!


Cheers

There is no need. I explained previously how to hear the problem. Create a snare hit with a velocity over 96 followed by a snare hit of 53, about a beat or less apart. Now try the same but make the softer hit a velocity of 51. The problem is pretty obvious. Takes about 1 minute.

That being said I soon will post an SFV for those interested.

It is, again, NOT an issue of the sample quality.

Dude!


you started the thread last night.   last night I checked out all the factory content, before I fell asleep.  I know what you are talking about.  I hate to break it to you, I do not think, "fixing the sfz is going to change anything."


So fix the sfz.  I don't know how to do it.  Upload the file.  Let us check it out.  

:-)


I believe the kit was not sampled well.  Steven's kits may sound great in his proprietary player.  But I am willing to bet they are only 4 layer kits.  caveat emptor

My earlier post clearly demonstrated "fixing the SFZ changed something". Significantly. The jpg labed Test shows this to be true. What's not to understand?

Thanks Albert.

The hits in the video are center hit articulations only... there's a separate articulation for Rimshot, Drag & Sidestick.

Cheers,

Rail

Same source as you with some trial and error and experimentation... and a direct email to the source if I have any questions.

I obviously can't comment on what others may do.. but we spend a lot of time and effort trying to make our libraries the best they can be....  I do disagree with the poster who thinks it's all about getting the mics up and hitting record...  there's a lot more to it than that and it takes us a long time and a lot of $$$ to create one expansion library...  about 6 figures.  

I do know that no other company has the drummer hit the drums as many times as we have them do it though (they all love me )   

Cheers,

Rail
-----------------------
www.platinumsamples.com

If we put the drummer's name on the box they hit the drums....  I have Chad Smith of tape/video saying "You got to be %^$$^% kidding me" when I told him to hit the drum 300 more times 

Each drummer has a distinctive way they hit the drums...  we couldn't get Chad's sound without having him hit the drum. 

You are correct however.. it's in the semantics of how the kit is marketed... does it say that XYZ hit the drums or is it just their kit?  Having Joe Blow hit the kit doesn't make it sound the same as when XYZ plays the kit... it's all in the wrist.   

Rail