How can I normalize entire instruments?

[gabriels]

I\'d like to normalize all samples in several instruments. I have both SampleWrench, full version, and SoundForge, but can\'t figure out how I could easily extract all wavs from a gigainstrument, normalize the samples, and then stick them back into the .gig. Can someone suggest how to do this... either with the two tools I mentioned, or with something else?
Gabriel

[imperium]

Hi,

Why would you do this? Normalization can be very destructive for you audio. The quantization noise won\'t get any softer (louder instead) by normalizing your samples. Have they been recorded really soft?

Barnier

[Chadwick]

Gigastudio allows you to assign a wave editor as the \'preferred\' sample editor. At this point I believe it works only on a per sample basis, not on batches.

To work on groups of samples you can export a whole folder and deal with them. I think as long as you bring them back into the editor with the same names, your instrument will remain correctly mapped.

For indiviudal sample edits:

1. Open the editor
2. On the menu bar choose \'Edit\'
3. Under Edit choose \'preferences\'
4. On the right hand side of the window under \'audio editor\' insert the location of your preferred editor

Once this is done, from now on you can right click on a sample in the sample pool, choose \'edit audio\' and have the sample copied to your editor. It will probably appear as something like \'GS000000.wav\' so keep you wits about you. You don\'t need to rename as Gigastudio knows that this temp file is referenced to the original sample. If you screw up the edit - just don\'t save it.

If you just want to keep the original instrument and samples as a safety, copy the instrument, right click on a folder in the samples pool and choose to \'export all sample folders\'. Now you can batch edit to your heart\'s content.

[Bruce A. Richardson]

Originally posted by imperium:
Hi,

Why would you do this? Normalization can be very destructive for you audio. The quantization noise won\'t get any softer (louder instead) by normalizing your samples. Have they been recorded really soft?

Barnier <font size=\"2\" face=\"Verdana, Arial\">Normalization gives certain instrument design advantages, although the cleaner way to do the job is to raise levels during the session for softer attacks and compensate during the mapping stage.

Especially for people who use wind-controllers, one needs a method for making all samples of various dynamics the same recorded volume. Otherwise, the wind control essentially provides a double attenuation.

[imperium]

Right. That makes sense. images/icons/smile.gif

As sounddesigner/audiopost engineer I always jump off my chair when normalization comes up. I\'d better tie myself down while surfing this board. images/icons/grin.gif

cheers

Barnier

[gabriels]

[gabriels]

Well, my situation is that I _do_ use a wind controller, and that I have a library that has been recorded at quite a low level.

I believe this was done because it is a library of solo instruments, and so the instruments are \"supposed\" to be relatively less loud than, say, a string section. This might work in some situations, but I\'d prefer to have the option of using the samples at \"full\" amplitude in some situations, and then attenuating them in others.

The problem is that I\'m simultaneously using other giga instruments that were recorded at much higher levels. Of course I could adjust the mix in the Giga DSP, but I\'m using other synths and samplers as well, and I hate to start out with a mix that is attenuated and then boost it.

Of course, that\'s kind of what I\'d be doing to the library in question, but at least I wouldn\'t be doing it to all elements of my mix.

I\'m wondering, however, if normalizing will rob the samples of some of their variability. Obviously, with a wind controller, I can vary the amplitude of notes as I like, but I\'m just a little nervous that, by normalizing, I\'ll end up with a somewhat more mechanical sound. Has anyone any experience with this?

Yes, Chadwick, that\'s how I\'ve started doing the normalizations, but there are so many samples that it\'s really tedious. I was hoping for some sort of batch ability. I\'ve heard that Awave might do this. I\'ll have to check it out. Then again, maybe I\'m about to change a pretty gorgeous, but too softly recorded, library into a monolithic piece of junk.
Gabriel

[gabriels]

But will CoolEdit help me in the process of extracting the wavs from the gigainstruments, and replacing the samples into the gigainstruments when I\'m done normalizing.

Seems as if you\'re saying that CoolEdit will somehow keep the relative differences in amplitude between the samples. This would be cool indeed!
Gabriel

[KingIdiot]

gabriels, it depends on the patchess you want to convert.

I\'m guessing its Dan Dean Woodwinds.

The samples in this lib are compressed into a giga native format that allows for better streaming possibillities and less HD space.

they are also not exportable via the Gigaeditor. Which is where your problem may be.

For instance, uncompressed samples in a gig file can be exported to a folder of your choice, then you could batch process them in another application, then \"replace\" them via the giga editor.

However since you cant export compressed files, you have a problem.

AWAVE Studio can export compressed files, and even normalize files within the gig instrument for you. Great tool, only problem is it doesn\'t support giga 2.5 modifications. So if there are any 2.5 specific programming features, you\'ll lose them, not to mention it destroys the folder tree within the gig instrument.

Still for DDWW instruments it should be fine, since I dont remember it having much in the realm of 2.5 specific programming.


otherwise you could use an app I befriended a year or so ago, named Quick Keys. You could possibly build a single keystroke command that would open a file in sound forge through the gigaeditor, then normalize, then save/close, and move to teh next file and repeat.

Thus a \"auto batch\" creation of your own.

Another big King Idiot Secret out in the open.....

believe me if you put your mind to it, you can make quick Keys do SO MANY things.

[gabriels]

King,
Actually, it\'s the \"pre-release\" of KH Solo Strings that I\'m trying to normalize. Great sounds... very usable for my purposes, but recorded at a low level... possibly to fit in as elements of a larger mix.

For me, it would be better to have more gain in the samples.

The samples are extractable to a folder. I have been using SoundForge to do the normalizing on some of the samples. It\'s a little tedious and quick keys sounds as if it could be a help. But one of the most tedious aspects of this process is putting the edited wavs back into the gigainstrument.

I think I understand, or I should say, I don\'t REALLY understand the structure of a gigainstrument. It seems to me that the samples are in a different format when they are stored in a gigainstrument. They are extractable as wavs. If I edit the wavs, I\'ve found that I still need to replace the individual samples in a giga instrument with the edited wavs. Very tedious indeed. Am I on the right track? Sometimes I think I am, but that I\'m headed in the wrong direction.
Gabriel

[Simon Ravn]

Normalization would definitely be a plus for libraries like London Orchestral Percussion. Some of the stuff there is so low you wouldn\'t be believe it - and it makes the transitions between velocity layers very unrealistic in for example the toms and glockenspiels. I checked the .GIG\'s and everything is as loud as it can be with the samples that are in the patches. So the only choice to fix these issues is either to 1) Normalize the samples by hand, then attenuate them properly, or 2) Make the loudest samples more quiet - which would bring the general volume of many of the programs down where it wouldn\'t be very useful.

[gabriels]

Oh, I forgot to ask.... which programming features are specific to 2.5, in case I decide to try Cool Edit Pro?
Gabriel

[gabriels]

King,
I also forgot to ask what you meant by destroying the folder tree within the giga instrument.

Let\'s see... my giga nomenclature is failing me now...

I can have a giga something which has several instruments inside it..

So, I might have \"symphonic strings\" which has
\"cellos\" \"violins\" \"violas\" and \"basses\" as instruments. Isn\'t \"symphonic strings\" a giga instrument. Then what are the \"violins\" etc. Anyway, I could figure that out from sitting in front of my giga pc.

But the question really is, will I lose the internal structure of \"symphonic strings\" which allows me to see violas, basses etc as separate instruments within the gig?
Gabriel

[Ray Lindsley]

Wave Lab has a function called Metanormalize that allows you to batch process multiple audio files and normalize them to each other. It allows you many options for saving the resulting files including replacing the original files with the process files.

[Chadwick]

Normalising won\'t change the timbre of an individual sample. However, if you are going to normalise so that each individual sample is at its maximum volume, you may need to take note of the relative amplitude differences between the samples before you normalise, and then reapply that level difference in the editor later.

Donnie (UOP, LOP etc.,) once pointed out to me that the harmonic makeup of a sample may make it sound exactly right level-wise relative to its neighbours even though it is a few db softer. If this sample and its neighbours are then normalised, it may suddenly sound \'too loud\'. You just need to keep an eye on the relative levels if the overall instrument is sounding well balanced before you normalise.

The big difference between right clicking on a single sample and choosing \'edit audio\' and exporting a folder full of samples is that only the single sample edit updates the sample in Giga automatically. Wavs edited in an exported folder need to \'replace\' the originals one by one. Hopefully V3 will sort this.

As King said, two different types of wav files are used by Gstudio. One is the normal pcm wav, the other is a compressed pcm wav which loads and streams much more efficiently with zero quality loss. Compressed wavs are incompatable with the export functions.

[KingIdiot]

yah but \"metanormalizing\" or normalizing across all files toa relative peak, not peak normalizing each files, could be a good practice if you have many velocities and want to get every last DB out of the samples.

also thats another issue. peak normalizing isnt \"percieved loudness\" thats what Donnie\'s talking about. Its more noticable in sustain notes, rather than short percussive bursts if you ask me.

Still room harmonics can create obscene resonance and make you want to adjust levels of particular samples for the sake of consistancy. It all comes down to the particulars of the way you want to program a library if you ask me.

There are other advantages to normalizing when programming files. Like with Xfades for instance. You can avoid most shelving artifacts, especially with Giga\'s Linear Fades.

You can also program more \"playable\" velocity transitions with volume adjustments (and that is usually easier with Normalizing).

I think the idea of keeping things at natural recording levels isn\'t the best option if you ask me. Unless you\'re doing 16+ velocity dynamic samples images/icons/smile.gif Samples ARE NOT real instruments. So I dont believe their sample recordings should be treated with \"real world\" sound physics. Of course it really depends on your recording AND instrument/sample library strategy. You may not depend on velocity switching as much as other libs so it wont matter as much, or you may be recording with multiple mics or far mics to which you\'d want to keep a consistant ambience level. Those issues make for not wanting to peak normalize each file.

I do agree with Bruce tho, the best way is to compesate at the input stage then attenuate for best bit depth and S/N ratio, but obviously thats tedious

record at 24bit and you should be fine and \"normal\" volumes, unless you\'ve got a ton of room noise....which then you gotta be careful with. Normalizing will give you fluctuating room sound. (in which case, the devil of noise reduction can help, used in moderation, or supply a general \"noise mask\" sample that we can use to NR to our own taste, and also be able to use as an \"overlay\" sample to add natural noise to the mix)

all sorts of issues to deal with and ways to work around them images/icons/smile.gif ... not to mention perogotives(sp? where\'s boby bown when you need him?)


gabriels,

what you can do is get an application that will batch process. Cool Edit, Wavelab, AWAVE, all do that, and then there\'s sonic foundry\'s nightmare \"batch processor\" an actual seperate app from sound forge (how stupid was that to remove it from SF4.5?)


What you need to do, is export the samples to a folder, then have one of the other applications \"batch process\" all files in the folder, normalizing either at relative peak (keeping differences in volumes the same), or at individual peaks (all will be as loud as possible without clipping). Then in the giga editor, you right click on the folder you exported and select \"replace all in folder\", then select the folder witht he processed files.

Save, and all should be fine....so long as you dont run into unity note bug/issues with giga retuning stuff for you images/icons/smile.gif

I know there\'s a way to make giga ignore this stuff, I just forget how to do it at the moment.

anyhow, good luck. Withwind controllers I suggest individual peak normalization, and then working to attenuate settings to your liking.....

not fun, but the best option for sound.


BTW all of the above is just stuff I\'ve seen, not really a matter of fact since its all opinion anyway.

[gabriels]

REPLACE ALL IN FOLDER????!!!!
Why didn\'t I see that option? Thanks. Can\'t wait to find the button and click it. I\'ve got SoundForge 4.5 but have yet to figure out how to batch processes with it. I\'ll check out the help file.

The main thing I\'ve worried about in normalizing is screwing up looped waveforms.... Yeah, I know Gigs are mostly unlooped, but some aren\'t. I suppose if the normalizing is done correctly there should be no problem, but I seem to remember creating some artifacts in loops for my EMU E6400 using SoundForge. I suppose I\'ll find out about this soon enough.
Gabriel
Gabriel

[KingIdiot]

hmm hard to say. I dont think you\'ll lose loop data if the file stays the same format (wav) and you only normalize.

I remember losing loop info when converting formats (wav-aiff/aiff-wav). But all should be good.

and the batch processor in 4.5 should take care of you. now RTFM images/icons/wink.gif

[gabriels]

it\'s not loosing the loop info that I\'m worried about, but if the normalization isn\'t just right (which I\'d think it would always be), then the splices might go slightly out of whack.... what was originally a seamless splice becomes audible. That\'s what I remember. Antares Infinity helped me out of that hole. Pretty amazing product for getting rid of problems in loops.

RTFM.... Hmmmm.
Return to Frequency Modulation?
Repulsive to Female Musicians?
RotoTom Flextone Madness?
Rolodex Tamarind Fandango Matador?


Gabriel

[KingIdiot]

if the loop is seamless in the first place, it should be fine after normalization.

Its when offsets get screwed up that you loose the seamlessness.

if it DOES happen (not likely)

usually a simple high resolution sample level Xfade will take care of it and not be audible.


jeez...I actually sound like I know what the hell I\'m talking about in this thread....

[thesoundsmith]

Cool Edit can do this - give it a list of samples, and it will go through them all and find the max volume, and use that as a reference to normalize the entire group relative to the3 loudest (if the loudest were -3dB, they\'d all be made 3dB louder.) I assume (but haven\'t seen it done) that you could also just batch them and have each sample nomalized to 0dB (or whatever)

Dasher

[gabriels]

Agreed!
But, you\'re not going to leave me hanging about the meaning of RTFM are you?
Gabriel

[gabriels]

Oh, I just did a Google search on RTFM
Blush.... shame....hanging head....feelings of inadequacy...
but, wait a minute, I\'ve read enough manuals to earn the right to, now and again, plead ignorance and ask for divine intervention.
Oh, all right, I\'ll read it.
Gabriel