KVR Audio

sinkmusic · Post by **sinkmusic** » Mon Oct 31, 2005 4:42 pm

Hello
I have a bunch of audio files i would like to put at the same level.
So I would like a free transparent plugin that could be use for levelling without making too many artifacts.
I don't think that a maximizer (like buzroom Buzmaxi3) suits my needs, i'm looking for one that could analyze the average level of the given audio file and boost its level until a specified level.
Isn't RMS compressors aimed at that ?

Do you know any freeware plugin i could run in a batch job in SoundForge ?
Thank you.

robenestobenz · Post by **robenestobenz** » Mon Oct 31, 2005 8:10 pm

Doesn't Soundforge have a normalisation feature?

sinkmusic · Post by **sinkmusic** » Mon Oct 31, 2005 9:00 pm

Yes, it does have, but when i used it, it acted too much like a huge compressor.
I just want something analyzing the average level of a file, and making it louder for a specified amount without compressing it (or just the peaks).
It is for one shot samples. I have plenty of them, but some kicks are very loud, and some hats are very quiet. So i would like to make them sound at an average level.
I can't remember that SF could do this, his normalisation feature was not transparent, it was overblasting the sound (but i did not use from years...)

sinkmusic · Post by **sinkmusic** » Mon Oct 31, 2005 9:05 pm

Well, that's right the topic's subject should have been "auto-levelling" plugin : something putting a bunch of samples all at -3db, or -6,etc. without compressing them...

bduffy · Post by **bduffy** » Mon Oct 31, 2005 9:13 pm

You should be able to get a pretty straight-foward Normalization in SF. If you use the "Maximize Peak Values" and switch it to RMS detection it defaults to "apply dynamic compression". Switch it to "Normalize Peak to 0db" and you might get something less harsh.

greendoor · Post by **greendoor** » Mon Oct 31, 2005 10:29 pm

Since this is for one-shot drum samples, and not for music, I don't believe you should compress or normalise them at all. You will be setting the level in your mix, and maybe compressing and limiting. Don't damage these samples, keep them pure.

sinkmusic · Post by **sinkmusic** » Mon Oct 31, 2005 10:49 pm

Yes, but it is for drumkits sounds, and i want each kit to be at the same level.
thanks for your advice.

smart · Post by **smart** » Mon Oct 31, 2005 11:02 pm

LUXONIX LFX-1310 has a RMS compressor built into it.

bduffy · Post by **bduffy** » Mon Oct 31, 2005 11:12 pm

If you have SF8, you can use the Batch Tool to set up a Batch Job (or BJ

) and batch normalize them.

I kind of agree with Greendoor; these days I tend to use compressors/EQ/Saturation to maintain gain in drumkits, but I've made some great drumkits using offline processing too, so whatever wets your beak.

greendoor · Post by **greendoor** » Mon Oct 31, 2005 11:14 pm

sinkmusic wrote:Yes, but it is for drumkits sounds, and i want each kit to be at the same level.
thanks for your advice.

Do you mean you are designing drumkits for a sampler or drum sample player? When you build your drum kit patches, you have to set the level and pan for each drum. I see no advantage in what you are trying to do.

As soon as you start using your drumkits in a mix, you will want to adjust the levels of each drum anyway.

robenestobenz · Post by **robenestobenz** » Tue Nov 01, 2005 12:46 am

greendoor wrote:When you build your drum kit patches, you have to set the level and pan for each drum. I see no advantage in what you are trying to do.

I can see one. I think it's better to let the sampler determine the volume rather than the sample's recorded level. Otherwise, variations in the volume of the sample coupled with a sampler's usual response to lower velocities can make for inconsistent volume between layers.

MGB · Post by **MGB** » Tue Nov 01, 2005 1:13 am

http://www.plompy.co.uk/software/wave2_11.zip

wave file leveller (feeware)

greendoor · Post by **greendoor** » Tue Nov 01, 2005 3:23 am

I think it's better to let the sampler determine the volume rather than the sample's recorded level. Otherwise, variations in the volume of the sample coupled with a sampler's usual response to lower velocities can make for inconsistent volume between layers.

You mean when creating velocity layered drumkits - where you are also using velocity sensitivity for level as well? I know that some developers do that, but I would respectfully suggest that its not a good way to design a drumkit. For maximum realism, I believe it is best to use the samples at their natural level, and use velocity switching alone to choose the sample that plays. If you smash the dynamics of every sample so that they are all equally loud, and then use velocity to change the level, it's not really realistic. Personally, I would prefer to use just one sample and just velocity filter it if you are going to tamper with the sounds that much.

Just because a sampler or drum player offers several choices of velocity control, it doesn't mean you have to use all of them.

robenestobenz · Post by **robenestobenz** » Tue Nov 01, 2005 3:47 am

greendoor wrote:You mean when creating velocity layered drumkits - where you are also using velocity sensitivity for level as well?

The velocity sensitivity would determine the volume exclusively, and would trigger different kits in the normal fashion.

greendoor wrote:For maximum realism, I believe it is best to use the samples at their natural level, and use velocity switching alone to choose the sample that plays. If you smash the dynamics of every sample so that they are all equally loud, and then use velocity to change the level, it's not really realistic. Personally, I would prefer to use just one sample and just velocity filter it if you are going to tamper with the sounds that much.

Sure, I wasn't arguing the "right way" of doing things - just saying it suits some people. Though it's a funny thing you say there at the bottom seeing how the word realism lingers not far above. Seems to me that velocity splits would quite obviously be the biggest factor for realism, with either method of determining level.

aciddose · Post by **aciddose** » Tue Nov 01, 2005 4:04 am

what you need is to use a sample format that supports a floating point gain value, and then normalize (range so maximum value of the sample is equal to the maximum value represented by the sample format). then you need to go through either by hand, or calculate the rms of each sample and adjust the gain value to equal 1/rms.
calulate rms as follows: go through the entire sample and sum sample^2. the rms = sqrt(sum / samples)
that is: root: sqrt, mean: average, squared: sample^2.

if you dont know of any formats which support a float gain value, dont worry, neither do i. why the hell didnt someone implement this in the first place? because they're f**king morons, if you wonder.

freeware RMS compressor ??