News:

2box forum: accident-free since the last one.

Main Menu

Dilemma of making DSNDs

Started by espen, February 05, 2012, 04:16:37 PM

Previous topic - Next topic

espen

Hi.
I've spent hours and hours for solving this problem but I couldn't make it. We currently have two ways to create DSND files but neither is working optimally.

Dsoundtool: First, I couldn't have solved the memory problem althought I'm very aware of the RAM increasing procedure. Doesn't work for me. I must say that the problem comes only when trying to convert larger than approx. 150 MBit wav files. Secondly I noticed that there is some kind of threshold value in the Dsoundtool that ignores the quietest hits. It just say "not detectable hits" or something like that.

Sound Editor: None of the earlier are problems in Sound Editor and I would like to use it now as it supports multi zones etc. but there the problem is that it creates very annoying hiss to every sample. Because of this it's unusable.

Has anyone faced similar problems? Is there some solution for these? Especially, I would like to hear if you have used Sound Editor succesfully.

edtc

do you use 16 or 24 bit wav files when you get hiss ?

espen

#2
Both. Have you noticed something between those?

Louis

Hi Espen,

Have you tried the "One directory containing many wave files" option in DSoundTool when generating the dsnd file?

Regarding your memory problems with DSoundTool, what OS do you use? How exactly do you start the tool?

espen

Hi Louis.

And thanks for your work for the 2Box community.
So far I've made one wav file which include all the hits. Then for hihat and ride especially, I've splitted the file and made the final dsnd from separate files. That way I have managed to make e.g., about 500 MBit hihats.

I start the tool usually from the jar file but I've noticed that the memory is a bit better when launched from web site. My OS is Win7.

Is there a possibility to lower the threshold that I mentioned in the first post?

Louis

Hi Espen,

First the memory: if you download and unzip http://www.shortestpath.se/dsoundtool/dsoundtool.zip you will find a file called dsoundtool.bat. You can start DSoundTool by double-clicking on the bat file, which will give the tool max 1.3 GB memory. If that is not enough, read the instructions in section "Increase memory usage" on http://www.shortestpath.se/dsoundtool/download.html

Second, the threshold. Yes, you are right, if you use 1 sound file containing all samples, DSoundTool uses a threshold value which is very low and skips samples with max volume below that threshold. This is mostly needed to skip noise in home-recorded samples. Unfortunately it is not possible to change this value. But if you split the wave file using some other program like SoundForge, then you can use the many files per directory option. In that case DSoundTool will take all samples as they are, and will not use any threshold.

Hope that helps you...

edtc

#6
In soundforge ,there is a tool named "Autoregion"... after playing with parameters ,i found a way to have a reliable preset  that worked with a kick that Dsoundtool didnt wanted ....

here is how i set parameters for autoregion :



By the way Louis , welcome back !!

after autoregion , you have the option to extract those regions .... soundforge adds a number from 1 ( first region) to 30 if there are 30 regions  and then splits the file in the folder you chose ...

In DsoundTool , could it be possible to have the option , in multiple file case , that the number after the sample's name determines its position... i.e.  : sound1.wav is first  layer , even if it s not the loudest ...?


Manfred

Quote from: edtc on February 06, 2012, 09:35:45 PM

In DsoundTool , could it be possible to have the option , in multiple file case , that the number after the sample's name determines its position... i.e.  : sound1.wav is first  layer , even if it s not the loudest ...?


Hello,

this is possible, maybe not with file names but with a mapping file. Have a look here:

http://www.shortestpath.se/dsoundtool/manual/advanced.html

There you can set the velocity for each single file. Never tried by myself but saw it at the manual.

Regards, Manfred

edtc

Quote from: Manfred on February 06, 2012, 10:10:53 PM
Hello,

this is possible, maybe not with file names but with a mapping file. Have a look here:

http://www.shortestpath.se/dsoundtool/manual/advanced.html

There you can set the velocity for each single file. Never tried by myself but saw it at the manual.

Regards, Manfred

... Ja wohl , i knew this possibility ,but i thouht it could be quicker to do it this way in some cases...  especially when you prepare the files in soundforge , and want a special order for the files ...

editing a txt file is long and boring...

espen

#9
Quote from: Louis on February 06, 2012, 06:51:12 PM
Hi Espen,

First the memory: if you download and unzip http://www.shortestpath.se/dsoundtool/dsoundtool.zip you will find a file called dsoundtool.bat. You can start DSoundTool by double-clicking on the bat file, which will give the tool max 1.3 GB memory. If that is not enough, read the instructions in section "Increase memory usage" on http://www.shortestpath.se/dsoundtool/download.html

I have a couple of questions. My problem is that when over 1500MBit the tool doesn't start from bat file. It gives some java error. Could the problem be linked with that?

Should this memory increasing procedure effect when launched from website, or does it matter?

Quote from: Louis on February 06, 2012, 06:51:12 PM
Second, the threshold. Yes, you are right, if you use 1 sound file containing all samples, DSoundTool uses a threshold value which is very low and skips samples with max volume below that threshold. This is mostly needed to skip noise in home-recorded samples. Unfortunately it is not possible to change this value. But if you split the wave file using some other program like SoundForge, then you can use the many files per directory option. In that case DSoundTool will take all samples as they are, and will not use any threshold.

Hope that helps you...

Thanks for this tip. Maybe I will try it or I will record the hits separately.
Thanks for your time!

ps. Oh, one more question. Why is the amount of samples in Dsoundtool limited to 99 hits per articulation? Actually, I don't know if more is needed but I noticed that there are some 2Box samples which have more than hundred samples.

Slap the drummer

Quote from: edtc on February 06, 2012, 11:11:08 PM

editing a txt file is long and boring...

It doesn't have to be too bad......   :)

From memory......
a) know what the file names need to be
b) know how many you've got

Use Copy/Paste and Find/Replace to make up the full text file by editing an already
existing mapping file.

Then you just need to whizz thru and assign the velocity values.  Depending on what you
want to achieve, I seem to remember using simple formulas in Excel to make EG the sequence
of values I wanted - then just copy/paste the numbers in......

Ah, just remembered - in fact I did all the editing in Excel, with a final column stringing all the results
together to make each line of the finished text file.
It worked really well, quick and hassle free  :)



Slap the drummer

On the memory issue.....  speaking as a total layman....

I think what happens is that the OS will refuse to accept the Java command to assign
memory allocation if it is "too" large.

So (running Vista) I kept editing the BAT file to find out what was the largest allocation
that didn't generate an error.  On mine, 749Mb is Ok, but 750Mb and the application won't
load without an error.

I googled this a bit and it seems like there's a lot of people out there wanting to know how
to make Java claim the necessary memory....

But none of them were drummers and it all looked very complicated  :(

Louis

Quote from: espen on February 07, 2012, 06:07:37 AM
ps. Oh, one more question. Why is the amount of samples in Dsoundtool limited to 99 hits per articulation? Actually, I don't know if more is needed but I noticed that there are some 2Box samples which have more than hundred samples.

This for historic reasons only. There is no specification of the dsnd file format so I had to reengineer it. At that time there were no samples with more than 99 samples, and at some place in the file the number of samples is stored as a decimal number. I did not (and still do not) fully understand the dsnd file format and I did not dare to produce dsnd files that contained 3 decimal digits that perhaps would not play in the module, so I decided to play safe.

It would involve some work for me to increase this number, because I would need to find out how such sample numbers are encoded. If you can point out a dsnd file with 99 samples I can have a look and see if it is just an easy fix (but no promises). But honestly I don't think anyone will hear a difference between a dsnd file with 127 samples or 99 samples.

Louis

Quote from: edtc on February 06, 2012, 09:35:45 PM

By the way Louis , welcome back !!

after autoregion , you have the option to extract those regions .... soundforge adds a number from 1 ( first region) to 30 if there are 30 regions  and then splits the file in the folder you chose ...

In DsoundTool , could it be possible to have the option , in multiple file case , that the number after the sample's name determines its position... i.e.  : sound1.wav is first  layer , even if it s not the loudest ...?

Hi edtc, nice to talk with you again! Your proposal is a bit difficult to implement, because I do not know what velocity to assign to each sample. I think Slap's Excel approach should be ok (this is anyway advanced usage). But hopefully some day I will realize your idea to make it possible to extend the sample dialog, you know the view where you see the individual samples, and in that view be able to delete individual samples, change their velocity, and perhaps their volume as well.

The problem is time. I spend too much time programming already, and short term I see no end in sight. That's why drumming is such a great pastime, just hitting things and emptying the mind... Hopefully before summer.

Jman

#14
Quote from: Louis on February 07, 2012, 08:57:47 PM
But honestly I don't think anyone will hear a difference between a dsnd file with 127 samples or 99 samples.

Not only will you not hear a difference I think 127 samples will be overkill. I started out with the philosophy that I could use 99 hits making dsnds with DsoundTool so that would be the way to go .... Bigger is better, right? Well, I have come to the conclusion that is not always true .... First of all bigger is also bigger on the SD card, so you will take up space. Sure you have 32GB if you did the mod, but you will probably find like I have that I want one more set of toms, one more snare with a dryer or wetter preset, one more great HH with 7 transitions for both bow and edge. Plus the HUGE dsnds can eventually cause more latency IMO during playback. It has taken me quite a bit of time to go back and make many of my dsnds more compact. I have compacted all of the largest dsnds I made in the beginning and IMO I have lost absolutely nothing as far as end sound quality. And from what I understand most VST soundsets do not include anywhere near 99 unique hits per instrument articulation. NDK does in some of their files, but I think even if you use every other hit with some of those you will still notice no difference in the finished 2Box dsnd performance. I will still use 99 hits at times for a snare head when building a dsnd .... but for the most part much less on most instruments/articulations.
Just my take on things ... and I am still learning as I go ... J
I could tell you where to stick that piezo! :D ;)
http://stealthdrums.com/