Topic: Post Processing - What is the prefferred sequence?

[boojum]

I have been doing some simple post production work in Samplitude LE and wonder what is the correct sequence of steps.  I am doing the post whatever, normalizing and then down sampling from 24/48 to 16/44.1.  Is this how it should be done.  Over in Gearslutz they just hanged five guys who were normalizing files.  Sheesh!  They do not seem to like normalizing, as opposed to compressing which, they detest.  I assume they hang, draw and quarter you and then feed the pieces to the dogs for that crime.

How are you guys doing it in whatever editor??

Cheers   

[JackoRoses]

I have been doing some simple post production work in Samplitude LE and wonder what is the correct sequence of steps.  I am doing the post whatever, normalizing and then down sampling from 24/48 to 16/44.1.  Is this how it should be done.  Over in Gearslutz they just hanged five guys who were normalizing files.  Sheesh!  They do not seem to like normalizing, as opposed to compressing which, they detest.  I assume they hang, draw and quarter you and then feed the pieces to the dogs for that crime.

How are you guys doing it in whatever editor??

Cheers   

I leave everything in 24 bit for editing purposes and use wavlab with the 32b float option enabled.
I don't normalize and I don't compress. I did just recently add gain to about 30 secs to a show I recorded.
But yes you have the right idea, do all your work in the 24 bit realm and than dither/downsample.
You should not be doing anymore processing on the 16 bit files once you downsample.
Unless all you have to work with is 16 bit and that's different...

[F.O.Bean]

I add gain/save 24/48 files, then resample, then dither, then track, then flac

[boojum]

I add gain/save 24/48 files, then resample, then dither, then track, then flac

Were you a drill sergeant??   LOL  You break it down simply as they do in the Army.  Example:  Sergeant standing before his class holds up a wrench and says, "This is a wrench."  And that is how it is done.  No lie.

[Patrick]

I have been doing some simple post production work in Samplitude LE and wonder what is the correct sequence of steps.  I am doing the post whatever, normalizing and then down sampling from 24/48 to 16/44.1.  Is this how it should be done.  Over in Gearslutz they just hanged five guys who were normalizing files.  Sheesh!  They do not seem to like normalizing, as opposed to compressing which, they detest.  I assume they hang, draw and quarter you and then feed the pieces to the dogs for that crime.

How are you guys doing it in whatever editor??

Cheers   

While I am not quite as serious as some of the guys over at Gearslutz, I definitely don't like it when tapers normalize files in post production.  Normalizing brings every level up/down to match the given peak level.  This takes away any dynamics that can really make a recording sound *great* instead of just noisy, compressed crap.   

I will say that I did at one time normalize my shows, but that was before I could tell a good audience recording rather than something that just sounded LOUD like studio albums. 

[Brian Skalinder]

I definitely don't like it when tapers normalize files in post production.  Normalizing brings every level up/down to match the given peak level.  This takes away any dynamics that can really make a recording sound *great* instead of just noisy, compressed crap.

Depends on how one normalizes.  Peak normalization doesn't change the dynamics of the recording - it simply adds the same amount of gain to the entire waveform, preserving dynamics in the process.  The "bad things" normalization does:  raises the noise along with the music, but since normalizing raises both the same amount the signal-to -noise ratio doesn't change;  since most editing software performs edits in 32bfp (or sometimes 24-bit), normalizing requires dithering back down to 24- or 16-bit (but if one's doing any other editing, the point is moot...one would have to dither, anyway).

RMS normalization, as I understand it (haven't used it myself), works like compression in that it reduces dynamic range.

I don't understand the beef with (peak) normalizing.  I mean, sure...in an ideal (16-bit recording) world, we'd all record with our loudest two consecutive samples hitting 0 dBFS (never mind the advantages of recording 24-bit).  But practicality dictates we leave some headroom.  Amplifying through peak normalization, or by cranking the stereo a bit louder...which is better?  I wonder if for most people (those with mediocre playback, or portable playback, or car playback, etc.)...it's probably less destructive / sounds better to amplify with a good WAV editor than the playback gear.  And if one's editing anyway...normalizing at the end doesn't do any harm - one's already converted to 32bfp and going to dither back down to 24- or 16-bit, anyway.

All that said, if I normalize I generally leave it as the last function prior to resampling and dithering.

[F.O.Bean]

I tend not to normalize, but change gain instead. i used to normalize, but eyars ago I thought I read somewhere it was better to CHange Gain rather than normalize in Wavelab anyway. I have no clue where I read it, or if my brain is maing that up, or if there is even a valid point for it, but thats just what I do I think I do it because I can get very near 0db without actually normalizing to 0db. I think its more precise whereas with the normalize in WL 5 you can only normalize in 1db increments cant you? like you can normalize to 0db or -1db or -2db but not more precisely right you cant normalize to like 1.05 can you? I can when I change the gain

[boojum]

I am doing post in Samplitude.  It is my understanding that normalization raises the volume of the file by the same amount through the entire file up to what ever amount you wish.  In my case it would be to -2dB.  Compression, on the other hand, will raise the volume of the quiet parts more than the loud parts so when viewed in an editor it looks as if everything is at about the same volume, because it is.  This is not the case with normalization.  If this is wrong, someone please point out the error.

Thanks.

Cheers

[ghellquist]

Normalizing as many other effects in Samplitude can be done in two different ways:
- destructive, writing a new data file with the effect burned in the file
- virtual, the preferred way.

In virtual normalizing all that is done is setting one of the gains inside the program, no sound files are changed.

My chain of postprocessing is like this:
- normalize to 95% (that is -0.45dB)
    explanation, I always run 24 bits with plenty of headroom, peaks hitting -12dB or similar
- high pass filter, somewhere between 20 and 40 Hz on the channel
    explanation, no music down there anyway, only muck.
- limiter on master channel, set to -2dB
    explanation, always limit before exporting. Saves from digital overs that can occur due to any processing.
- sample rate conversion and dithering always last, but that comes automatically from SAM

To that can be added other effects. Perhaps cutting the recording up into songs/audience. Sometimes audience sound is higher that songs, simple thing to modify the relative volume for each part.

I am certain you will find SAM a really good tool for post.

Gunnar

[stantheman1976]

I use Adobe Audition and use Hard Limiting finding the loudest parts and adding what dB I need to get those parts to around -3dB.  So the louder parts are not overpowering and the lower parts get a slight boost.

[jmz93]

I do what some people consider sacrilige, I add a little compression!

I work in Sound Forge, always with 24-bit files, and with SF set to use floating point temporary files. First I go through the file, looking for audience noise, especially Mr. Enthusiastic Clapper nearby, and bring that stuff down. If I'm right beside Mr. Clapper, I run his claps through a very SLIGHT! use of Click and Crackle Remover. If done right, it really does only remove the claps or knock them down a long way. 
If your software lets you hear what is actually being removed while doing any processing, be ure to listen to it.

Then I decide if the low end needs to be rolled off a lttle, or in rare ases, boosted a little.  I have some EQ preets set to mirror the filter of my battery box, so if it removes too much bass, I can at least listen to a attempt at bringing the lows back. If doing any EQ, it's almost always a case o slightly lowering offending frequencies, not boosting! 

Then, I go through some sort of smooth all-purpose compression, nothing extreme. maybe a ratio of 1.5:1 or 2:1, threshold at -15 or -18.  Or, I just use Sound Forges Wave Hammer plugin, Smooth Compression option, and reduce the ratio a lot.  Sometimes I use it's Master for 16DB preset if I'm eventually burning to cdr.

The better recording I have in the first place, the less fiddling I do with it after.  If it's great to begin with, I do almost nothing.