KVR Audio

bburtin · Post by **bburtin** » Sun Jan 02, 2005 7:26 am

Here's my first attempt to create a sampled instrument from scratch:

http://www.burtin.net/conga/

It's a velocity-switched instrument made up of samples I recorded of my friend Brad playing the various sounds of the conga drum. Hope that some people can find it useful. A big whoppin' thanks goes out to Rene for making sfz available to us for free!

Have fun,

Boris

Lunch Money · Post by **Lunch Money** » Sun Jan 02, 2005 7:32 am

Wicked!

Haven't tried it out yet, but I wanted to thank you anyhow.

Greg

Cordelia · Post by **Cordelia** » Sun Jan 02, 2005 9:08 am

Thank you bburtin! Trying them out tonight. Happy new year.

maximilians1 · Post by **maximilians1** » Sun Jan 02, 2005 9:41 am

Thanks .those sound nice,

Cordelia · Post by **Cordelia** » Sun Jan 02, 2005 9:49 am

Hey bburtin,
I'm going to download again, but first time through I got only CongaHeel4, CongaToe4, and CongaTone1, 2, 3, and 4. The other files were empty. Maybe a corrupt download- I'll try again right now.
OK, just DLd again, same result.

M'Snah · Post by **M'Snah** » Sun Jan 02, 2005 10:12 am

Sascha Franck · Post by **Sascha Franck** » Sun Jan 02, 2005 10:22 am

Thank you.
Here's a DR-008 mapping:
http://home.arcor.de/s.franck/BradsCongaDR008.zip
You will have to onvert the OGG files before using it. I used dbpoweramp to convert them to 16bit/44.1kHz/mono format.
Not sure whether the velo mapping is good enough, but it seems OK. Should be a minor deal to make it suit anyone's needs.

Cordelia · Post by **Cordelia** » Sun Jan 02, 2005 11:37 am

HansM,
I opened the individual Ogg files in SoundForge and some were empty. There must have been a problem on my end. I DLd your soundfont, and now all is well. Thanks to bburtin, and thanks, HansM, for the sf2.

learjeff · Post by **learjeff** » Sun Jan 02, 2005 3:48 pm

Thanks, Brad! Can't wait to try 'em out!

bburtin · Post by **bburtin** » Sun Jan 02, 2005 6:57 pm

Glad to hear that people are finding the conga samples useful. The ogg files that sound "empty" are just really really quiet. This is particularly obvious in the "heel" and "toe" samples. I didn't do much processing on the sample data because I wanted the recording to accurately reflect what Brad was playing.

That said, there may be some room for normalization. Does anyone know of an app that loads a group of audio files (WAV or OGG) and does "album gain normalization"? So it would normalize the loudest file to 0dB and then increase the volumes of the rest of the samples by the same amount. MP3Gain does this, but I believe it's only for MP3 files.

Boris

Cordelia · Post by **Cordelia** » Sun Jan 02, 2005 7:16 pm

You know bburtin, the files were actually empty-blank grey in SF. It must have been a corrupt download, nobody else has had a problem. The samples sound good, thank you for sharing.

scuzzphut · Post by **scuzzphut** » Sun Jan 02, 2005 7:27 pm

thanks for this, bburtin

good of you to share

pljones · Post by **pljones** » Sun Jan 02, 2005 8:10 pm

Boris,

If you're using un-normalised samples in sfz format, you'll need to adjust the programming. You've done nothing to tell sfz that the samples are normalised.

You need to add

Code: Select all

amp_velcurve_nnn=1

to each velocity layer, where nnn is the value you use in the hivel opcode for the layer. This tells sfz not to attenuate the sample volume at the "top" of the velocity range for the layer. From what you said earlier, you might then have to re-adjust all the layers...

learjeff · Post by **learjeff** » Sun Jan 02, 2005 9:53 pm

Usually, you want all samples normalized. Actually, that's not quite correct. Ideally, you want to record all samples so they peak nice and high to get a good recording (as usual). Practically speaking, it's not necessary.

Regardless, soundfonts generally work best if, for the loudest note on the keyboard, all layers of that note peak near 0 dB. Moving across the notes on a given layer, they might all be at the same level or different depending on the instrument in question.

Normalizing them all tends to work pretty well, though. Hopefully you've recorded in 24 bits and then normalize, so you get good16-bit samples in all cases.

Sascha Franck · Post by **Sascha Franck** » Sun Jan 02, 2005 11:55 pm

learjeff wrote:Usually, you want all samples normalized. Actually, that's not quite correct. Ideally, you want to record all samples so they peak nice and high to get a good recording (as usual). Practically speaking, it's not necessary.

Well, that's just not only not necessary but even impossible... imagine sampling, say, a piano - would you adjust the recording level for all "pp" samples, each time a key is played? Defenitely not.

So, you would simply do things as in a normal recording: Set the recording level as hot as possible for the loudest notes and record all the rest with the same record level.
After all, I actually seem to prefer lower leveled samples for lower velocity zones - that way I don't have to fool with individual zone level adjustments myself.

However, the loudest sample of each group of samples should be as hot as possible for the usual obvious reasons.

Free multisampled conga in sfz format