Soundfont : Attacks and Releases

encore · Feb 8, 2003, 08:57 PM

Hi there. My first post here so I'll get right to the point. ;)

I installed the Kx-drivers (3528) today. It's not the first time I've used Kx, I tried it this summer as well (think I had 3520 then), a lot of things are updated and for the better. So are also the soundfont-part. But there's one thing I reacted on back then and react on now, which is the following : All instruments with hard attacks sounds more dull with the kx-drivers than the aps/creative drivers.

I went to find out why. I sampled a squarewave and made a soundfont with just that sample, the attack and release on the adsr-envelope was set to 0.001 s (which is normal). Then I played the soundfonts via Cubase and sampled the result.

The original sample has no attack-time at all, it's just right on. The sampled result of the soundfont however was an attack of approximately 7 ms. (fading-in) I also tried to release the note before the sample had reached its end, the releasetime from signal to zero was approximately 15 ms.

Now my question is, can this be fixed? Maybe even in the 3530-update? Or is there a reason for this fading-behaviour? I can't think of any good reason for it, I mean if you wanted smoother attack, you could do this by increasing the attack time, right? As it is now, you could say goodbye to snappy bassdrums etc.

Last but not least, I'd just like to say that the Kx-drivers rocks and Eugene, you're my new idol! ;) Keep up the good work!

/// encore

**Eugene Gavrilov** · Feb 9, 2003, 07:22 PM

yes, it seems that attack is not calculated correctly

I hope this will be fixed in 3530

/Eugene

encore · Mar 4, 2003, 03:05 PM

Yeah, I hoped that too. To my dissapointment, it hasn't.. in 3531 maybe?

To give you an example of what I mean, check out this mp3-clip, that I just threw togheter.

tr808_without_attack.mp3 (165 kb)

Hear the lack of attacks of the drums? It's most noticeable on the bassdrum and the rimshot.

Max M. · Mar 4, 2003, 03:41 PM

it hasn't.. in 3531 maybe? - well before it could be fixed Eugene needs to find what exactly is wrong. So it would be very usefull if you also make mp3 with correct attacks (original drivers or aps maybe?) and also info on soundfont you used would be appreciated too.. (link maybe?)

encore · Mar 4, 2003, 05:21 PM

Max M. : Well, anyone who had heard the Roland TR-808 drummachine (even via Rebirth RB338) will know that there should be alot more attack to the sounds. If you read my first post, you saw that I've found out that the attacks (and also the releases) envelopes of the sound are much longer than they should be.

This has been a problem with all kX-versions I think, atleast it was in 3520 (which was the first kX I tried), and I'm actually surprised that noone else noticed it. Don't you guys like punchy basedrums? =)

Btw, the soundfonts I've used were made by me, so they're not available anywhere, but I could make a soundfont with some fast-attacked sounds, and also an mp3 with both the original sample played, and the ame sound played like a soundfont via the kX-drivers (the next best thing since I'm not planning to install the Aps-drivers again, however, the Aps-drivers are really ideal when it comes to the soundfont part, they've got the hard attacks (if that's what you want of course) and also SF2.1 compability!)

**Eugene Gavrilov** · Mar 4, 2003, 07:06 PM

we are currently working on the 'attack and release' issue
at the moment, we've found out that different Creative software uses different formulas
that is, attack time calculated by the E-mu APS drivers differs from the 'original' Live one and from the 'audigy' one (however, the synth engine is the same for all the cards)

this problem needs an acceptable and flexible solution...

/Eugene

encore · Mar 5, 2003, 02:26 AM

Eugene : Okay, that must've been a surprise. I can also imagine the hassle you're having while measuring the attack and release of 3 unique drivers. Keep up the good work!

Anyway, I think that the most important choice must be that the attack is calculated to what value you're actually setting it to in your sf-editor, so the driver that comes the closest should be the ideal one. Setting 0.001 should practically be 0.000, instant 100% attack ;-) I've had good experience with the Aps-drivers so naturally I think of these drivers as the ideal driver.

Another feature I long for (that are embedded in my soundfonts) are the extra goodies in the SF2.1 format, the possibility to assign things like filter/volume etc where I want to. For example, my drum-machine-soundfonts have the filter assigned to the modulation-wheel, for a lot of fun when making music.

Getting those features back again would be really nice!

I can understand that these feature will probably take some time to implement and that you've got enough to do already (really) but I hope they will be in there one day.

bennetng · Aug 9, 2004, 12:39 PM

Here is a good reference

http://www.sccmusic.250x.com/sfont/s...stic/index.htm

In My opinion, "Attack-based velocity-to-filter" described in that article need not to be emulated beause SF2.1 spec allows user to map velocity to filter cutoff manually. Then all things follow Creative's new driver (252 or later for Live, 3xx for Audigy, 4xx for Audigy2) should be OK becasue they are the same. kX also has a short compulsory release time and this problem should be fixed.