Topic: Pre/Post Processing?

[Music is alive]

It's ironic this topic appears here, now,  since - as a newbie taper - this is something I've been thinking about a lot lately.

When folks do choose to do things post production, what are examples of things that are commonly done?  This is assuming the doer is a regular civilian, not a pro mixer/engineer.

Brian S - please forgive the newbie question, but this post greatly intrigues me:

This is why it's so important to get your levels set well while recording - it maximizes the audio relative to the noise floor.  Using the example above:  assuming the same noise floor at the time of recording, -77 dB, setting proper levels at the time of recording to say -1 dB would result in a difference between the audio and noise floor of 76 dB.  So setting your levels properly at the time of recording maximizes the audio relative to the noise floor.

You've put your finger right on - without breaking out in newbie detail - a Q that I've struggled with, which, succinctly, is:  what is the difference between getting your levels just right during recording and (assuming you didn't come out with a badly clipped show) boosting it in post production?  

I understand there is a difference, but if you could use a bit of patience and expand on exactly the meaning of the "audio" and "noise floor", and their relation, for those of us who don't explicitly understand such wording, you'd be helping me a good deal.

[TNJazz]

Brian's explanation is very concise and clear.  If it seems a little fuzzy to you still, just try it at home with your stereo and you'll hear exactly what the difference is with your own ears.

Put your mics up to one of the speakers and set your levels so it hits around -2.

Record a 30 second clip or so.

Then back down the gain so the levels hit at -12 or -15.

Record another 30 second clip.

Then throw your samples into an audio editor and make them exactly the same volume.

Then listen to them and see if you can hear a difference.  I'm betting you will.

[zowie]

So how many of you secretly chuckle and roll your eyes when you see the word "remastered" listed in a DIME torrent?  I know I do...

I chuckle out loud.  Verbosely.

As far as post-processing, I don't like doing anything too involved or crazy with what I record though I do delight in having them boosted a little bit in Wavelab to sound "fuller" and "cleaner".  I sent my 2006-01-28 recording to Kfrinkle and he did some jazzy jazz to it in Wavelab.  I got it back from him and was amazed at how much nicer the recording sounded.

Is that so wrong?

"If it sounds good, it is good."  - Duke Ellington.
Corrolary: As long as there remains an unprocessed master around to go back to.

[Digital Quality]

SNR goes up 6db for each added bit in a digital signal. A full 16 bit signal correlates to 96dB available SNR; 24 bits results in 144dB of available SNR. If you record at less that optimal levels, it's just like leaving off (msb) bits on your final signal. For each bit you don't take advantage of, subtract 6db SNR.

Scaling afterwards will scale the data signal and the noise signal equally but then it also adds some new noise signal. To get the highest possible SNR in your recording it's best to use up as much of your resolution as possible up front. It can not be recovered later, all you can do is make it worse..

Now, if it's so quiet that you can't listen to it or your playback is super noisy at high gain levels, by all means scale it up. If you are trying to recover the last few dB you'd be best to leave things alone. In between there your ears are the best judge, do what sounds best to you. I try to never do anything extra to my recordings because they sound best to me that way.

[zowie]

Scaling afterwards will scale the data signal and the noise signal equally but then it also adds some new noise signal.

Could you explain the basis for that statement?

[Digital Quality]

Requantization noise is added. The scaled words must be brought back into the 16 bit domain. The difference between the scaled number and it's 16 bit word is the new noise. It can be helped by adding another level of dithering but not eliminated.

[cleantone]

SNR goes up 6db for each added bit in a digital signal. A full 16 bit signal correlates to 96dB available SNR; 24 bits results in 144dB of available SNR. If you record at less that optimal levels, it's just like leaving off (msb) bits on your final signal. For each bit you don't take advantage of, subtract 6db SNR.

Scaling afterwards will scale the data signal and the noise signal equally but then it also adds some new noise signal. To get the highest possible SNR in your recording it's best to use up as much of your resolution as possible up front. It can not be recovered later, all you can do is make it worse..

True. To the best of my knowledge anyway. I just want to note or add that the noise floor can be room tone and/or inherent noise from any peice of gear in the chain. Which of course is why you typed "available" SNR and not just SNR. Thought that might be helpful info for some people.

[zowie]

Requantization noise is added. The scaled words must be brought back into the 16 bit domain. The difference between the scaled number and it's 16 bit word is the new noise. It can be helped by adding another level of dithering but not eliminated.

If we're talking about the same thing, I don't think of that as noise but as distortion.

[Digital Quality]

Same thing. Distortion is the addition of a noise signal. Noise comes from a lot of places, quantization, interferance, components (even resistors add noise), not even considering frequency response of the system. Everything in front of the mics is considered to be "signal", even room noise. 96dB is the theoretical limit just for signal to quantization noise I believe. All the other stuff is added on.

Another strange thing about quantization is that you will still have added noise even if you say, "I'll just multiply everything by 2.0, that way there are no more rounding errors." The rounding errors are eliminated that way but you get a new effect of skipping quantization levels in your new signal. This I don't totally understand yet but apparently it's also a problem with the D/A converter. The quantization skips present themselves as noise when converting back to analog.

[wbrisette]

For me, I sometimes do it, sometimes I don't. All depends on the venue and what the recording sounds like. But, I also always do multitrack, so usually when I eq a track, it's a particular microphone and once tossed into the mix, rarely makes huge differences in the sound. I will also point out that many years ago, I use to master recordings for a living at a recording studio/cassette duplication facility. I learned a long time ago not to mess around with the recordings too much. I also learned a long time ago that it is better to subtract rather than add things to recordings (need more bottom end? drop the top and mids a few dB). Anyhow, I rarely put any of this info on my recordings, but a lot of that has to do with how I have to use a multi-track editor in post to do my work.

Wayne

[Digital Quality]

On the nicer tools, do they do extra dithering for you when you process or is that all left up to user? I have a cheapo wavelab student version it's pretty basic in terms of features. If you were doing several things it would be best to do them all at once and only requantize at the end is that possible??

As for UA-5 meters, I have a suggestion. Take it to some shows you don't really care about and run it so hot it makes you want to cry. The light should be blinking fairly regularly during the loud parts of the show but not on every beat and not glowing. Try it a little hotter during the 2nd set, remember you don't care if you limit or not. Get a feel for different brightness levels that it blinks and it gets a lot easier. I guess the led just tics on at -3db or so, is that right?

Your goal should be to just clip a few samples from time to time during the show IMO. The UA-5 handles a few clips with no problem but over all the time will wreck it. Any other opinions on that?

[Brian]

i know what i'm doing(or at least like to think so at this point ) so i'm all for it

Most 2 track shows I do now I send through Waves REQ 6 > L3 in Nuendo 3.  I do not use normalization at all as I don't think it sounds as good as a mastering limiter.  I also don't smash the hell out of the recordings either.  The out ceiling is at -.03db and I set the threshold at no lower than -3db depending on the initial recording levels so that it only limits the loudest parts and transients of the recording.

I've been quite happy doing this.

[VA_TAPER]

as far as EQ'ing... that is something that i want nothing to do with for the previously mentioned reasons.... tim's system sounds different than ray's, which sound different from mine

Maybe it's just me, but I'm an EQ guy (if it needs it).  EQ'ing your system and listening at reference levels (75-85db) is important before EQ'ing your show.  I run digital room correction, so I am not only EQ'ing out speaker/amp issues in my playback system but also phase, standing wave and time alignment issues.  But if we're talking EQ seting up your system is a cinch.  

1. Get on eBay and get an EQ ($20-$???)
2. Go to ratshack and get a dB meter
3. Get a testdisc of tones and pink/white noise
4. Sit at your listening position and calibrate the frequencies as needed

Now you are set up like a true monitoring system and can adjust you recordings to ear.  

mmmatt posted earlier a portion of an email conversation we had on this topic here's a snippet of it that shows my stance on the "debate":

>      3.  Do (did) you do any post on your files other than conversion? Everything seems so well balanced vocals > music > crowd, that I'm wondering if that is straight of the pre or if you tweaked it.  Maybe Glen travels with a kick-ass engineer?  I hear very little "room" in this recording.  Do you pull out resonance frequencies, or was this room just real dead... lack of heavy bass helps, but still.

I do post work on the shows, minor stuff but it brings
out some of the qualities you mentioned; what allows
me to do the post work is the frontend gear, by
capturing a clean signal without harshness I can adjust
several parameters (minor eq, compression, etc.) with
out bringing out the issues these tools did with other
rigs I've owned.  I do eq out seriously resonant
frequencies, this is done mostly by ear now, I run an
HTPC to do mastering.  If I hear something
(frequency wise) to strong I'll cut it using Waves
LinEq I never boost or cut by more than 3db, but 3db
can make a huge difference.

I cascade compressors, instead of using 1 compressor
for 6db of squishage, I'll use 3 different compressors
each providing a mild 2db.  This allows different
attack and release profiles to thwart pumping and
noise, but it allows the gentle equalization of volume
so that the banter can be heard and it brings out the
nice reflective qualities in the hall.

I de-ess as well to compensate for the issues of
recording a compression horn tweeter.

I used to add harmonics to the mix too, for bringing
out the highs, but the tube pre has completely
eliminated the need to do that.





     Thanks for the help.  I'm sure your gear helps make the recording sound so good, but that is just the flavors you like.  I wasn't sure how you feel about sharing so I didn't post this to the thread if you didn't want to answer these questions.  If you would rather I will post and let everyone hear the answers!  Once again if you don't want to answer these questions, there are no hard feelings.
 Matt

 
well, that turned in to more than a quick answer!

feel free to ask any thing else, really I buck
convention on alot of what I do.  People look down on
mastering, but no one plays their recordings back on
18" woofers and horn tweeters at over 105db so that
just doesn't make sense to me. I tape and master for
my ear, which is a little phucked up due to 7 years of
marine corps infantry, but generally most people like
what I do, so I keep it going.  Also, incase you haven't
picked it up I'm not an accuracy type of guy, I could
care less if my recordings sound like the venue, unless the
venue sounds good.  I'm a euphonic guy, I want my
recordings to sound like you're listening to a vinyl
record of the concert played through a vintage tube
amp.  


 

[jeromejello]

i should possibly clarify, i do EQ if necessary, but just thru my playback.  my point is that i dont EQ the file that i seed b/c of the differences (and lets face it, i am barely holding on to the craft of taping, let alone a whole new skill of post production) that can be experienced across the board.  i find most people prefer to have the sound in as close to a 'as it was' form and provide thier own tweaking with their own equipment for thier own taste.

[VA_TAPER]

i should possibly clarify, i do EQ if necessary, but just thru my playback.  my point is that i dont EQ the file that i seed b/c of the differences (and lets face it, i am barely holding on to the craft of taping, let alone a whole new skill of post production) that can be experienced across the board.  i find most people prefer to have the sound in as close to a 'as it was' form and provide thier own tweaking with their own equipment for thier own taste.

I'm not trying to convince you to EQ your recordings either, everyone gets in to this hobby for their own reasons, love of music, etc.  Not everyone gets as crazy/anal with it as some of us do (i'm in the crazy/anal group), but my point is there is not 'as it was' form.  Move your mic stand 10' to the left and you've got a potentially totally different sound than where you set up at, run a different pre, run a different polar pattern, 16/24 bit, small/large diaphgram, schopes/oktava all can do WAY more to the recording than some corrective mastering. You're the taper, be empowered (if you want to); do what ever you want to the recording that you think sounds good, seed it that way.  If people don't like they can tape it themselves, or you can learn from the critizisms. 

Either way, do what makes you happy. 

Here's an mp3 of one of the shows I taped and did a little mastering work on:
http://www.archive.org/download/glen2005-08-30.dsd.flac/glen2005-08-30set2t11_vbr.mp3

check it out, let me know what you think, I always appreciate new critics, it helps me hone in on the sound.

peace, and keep spreadin' the tunes,

chris