CD Mastering: One seamless WAV file or several WAV files?

[warble]

Hey all,
I've seen some posts around here stating the use of one big WAV file with all the tracks for a new project - then getting this file into CD Mastering software. I have CD Architect and am just learning to use it. I also use Sound Forge on a regular basis.

Guess I'm just wondering what the purpose of one file would be as opposed to all the tracks individually? I know you can use CDA to specify track placement, etc.

I'm just venturing into this aspect and any ideas are welcome. I searched around here a little bit, but wasn't sure what to specifically search for regarding this.

Warren

[thundercage]

I normally approach a project as a whole. An EP for example, has normally about 4 songs. You would want those 4 songs to all be at a similar volume and frequency range when played track after track on the CD. So, you would load all 4 in together as a single WAV and apply the same volume (normalize). Then put the appropriate track markers and burn to CD.

[Blue Bear Sound]

Originally Posted by thundercage

I normally approach a project as a whole. An EP for example, has normally about 4 songs. You would want those 4 songs to all be at a similar volume and frequency range when played track after track on the CD. So, you would load all 4 in together as a single WAV and apply the same volume (normalize). Then put the appropriate track markers and burn to CD.

That has to be one of the least effective ways of handling that.... cookie-cutter approaches almost never produce acceptable results.

And normalization is not an effective method of increasing level - unless the material has absolutely no transients..........

[adam_in_audio]

I agree on the whole "cookie cutter" thing, I've made mistakes from that ending in musical and artistic horror stories.

The best thing I've found to do is get a good mix and send it to a mastering house, worth the money IMHO.

Adam

[thundercage]

Originally Posted by Blue Bear Sound

That has to be one of the least effective ways of handling that.... cookie-cutter approaches almost never produce acceptable results.

And normalization is not an effective method of increasing level - unless the material has absolutely no transients..........

That's a simplified version. Most engineers last step is to normalize on completion of the final mix before mastering. What I'm saying is that each track of the recording should have the same 'volume' when played back. If you want to call that 'cookie cutter' fine.

[Blue Bear Sound]

Originally Posted by thundercage

Most engineers last step is to normalize on completion of the final mix before mastering.

Sorry Mark, this isn't true... if any sort of limiting is done to bring up levels, then normalizing is completely unnecessary and degrades the audio by a digital generation. Further - especially if it's going to mastering, there's even less of a need because the ME will be determining the final level and if the tracks coming in are too hot leaving no room for adjustments, then the ME has to reduce their level anyway before doing anything. Best off simply mixing it appropriately and sending the raw mix to the Mastering House with no gain added to the raw mix.

In Mastering, the basic premise is "never degrade/compromise the signal unnecessarily" - that same ethic should apply in earlier stages too, IMO.

[thundercage]

Here's a passage from 'Mastering Audio' by Bob Katz. I would assume he's fairly well respected in the mastering audio business.

'The engineer selects all segments (songs), and the computer grinds away, searching for the highest peak on the album. Then the computer adjusts the level of all the material until the highest peak reaches 0dBFS. If all material is group normalized at once, this is not a serious esthetic problem, as long as all the songs have been raised or lowered by the same amount. But, it's also possible to select each song and normalize it individually, which is part of the esthetic mythology - it's a real no - no.'

Then he goes on to say what Blue Bear say's that you shouldn't normalize if it's going to mastering.... interesting... I guess I assumed that was done because I saw another engineer do it and I respected the quality of his work.

If you are the ME where or why would you normalize?

[ecktronic]

My view on this would be to just record and mix the tracks ensuring good levels. And leaving the master volume of the songs up to the mastering company. Isnt that part of their job? I mean it costs enough!
If you dont have good levels in recording stage, doesnt that mean when mastering, the levels have to be boosted dramatically which gives a poorer quality finished product?

[thundercage]

I think the whole issue is a mastering question and not a mixing question. That being said, what purpose does 'normalization' serve in a mastering environment, if it can only degrade the quality of the recording? Should it be avoided? What is it's purpose?

Digitally would modifying the volume of a recording (not normalizing) introduce the same undesireable signal to noise ratio problems?

I was under the impression that you would degrade the recording, if it was resampled. Is this not true? I didn't think normalizing or volume changes resampled the content.

[Blue Bear Sound]

ANY digital process (include a simple volume change) introduces *some* degradation due to the round-off error that occurs during the mathematical rendering of that process. The degradation, done several times is not likely to be noticeable, but considering that in mixing a song, many volume changes occur, and many DSP operations take place, the potential for degradation is higher.

The reason Normalization tends to be frowned upon is simply because it's often unnecessary... for one - it's nothing more than a signal boost/cut, so the noise floor will get affected.

And most people (especially novices) reach for it expecting it will raise their overall levels, which is true for signals without many transients.. but a single transient can limit the amount of gain that is achieved. The more effective way is using Limiting to reduce the transients and therefore re-couping the potential for more gain.

[kylen]

Originally Posted by warble

Hey all,
I've seen some posts around here stating the use of one big WAV file with all the tracks for a new project - then getting this file into CD Mastering software. I have CD Architect and am just learning to use it. I also use Sound Forge on a regular basis.

Guess I'm just wondering what the purpose of one file would be as opposed to all the tracks individually? I know you can use CDA to specify track placement, etc.
...

warble - since you have CD architect you'll notice that when you drag your wave files into the project it pieces them sequentially one after the other as clips with markers and red book 2 second gaps automatically for you. You also get a visual of one long wave file, each one with volume seperate handles, effect inserts for each song, and effects insert for the master bus.

Folks that don't use CDA or something like it can append wave files to each other and specify track markers and whatnot but unless the person knows how to set all the redbook stuff then it won't necessarily be redbook. You can use Sound Forge to lay out tracks like that but you'd have to manually insert gaps somehow, etc. Also since it's one huge wave file in Sound Forge all the volume levels better be set right before hand. That's kinda hard to do unless you have the songs grouped in a multitracker or something like CD Architect, MAGIX makes some useful inexpensive stuff for this too.

As far as normalizing goes this generally won't do the trick for a CD project but you should try it out to see why. Even the better ones that try to normalize based on rms levels will only get close - I think Sound Forge has one like this. It's a long story but there's a zillion threads around here on it if you want to research it.

Regardless, in CD architect you can simply begin playing the project and adjust volume handles to manually 'normalize' the project. Since the term 'normalize' has been demonized somehow perhaps the phrase to use is 'adjust loudness' or volume of all the tracks in the project. If you want to you could even stick a mastering limiter on the master bus so a few of the peaks are pushed down a dB or 3 - easy does it though - not too much! I'd adjust all the volume handles first though before trying to get things to sound HOT. There's a free VST mastering limiter called Classic master limiter oddly enough:
http://www.kjaerhusaudio.com/classic-master-limiter.php

I assume you're trying to do this yourself, that's what I do at the home project studio level. Anyway once you get your project on CD then play it in a few systems and see if the volume levels are set right meaning you don't have to reach for the volume knob on any of the songs - except the first one of course!

2 cents

[VU-1]

It doesn't make any difference at all whether you place your mixes on the EDL (Edit Decision List - fancy term for 'Timeline'...) as individual wav files or as one large wav file. Even if you are using one large file, you can still break it up into regions or objects or whatever your program calls them and shift them around however you want them to appear on the finished product. I can tell you this: an ME worth his salt will process each song individually anyway so you might as well mix them to seperate files...

As for normalizing:
I never normalize. I use a limiter to obtain the desired RMS level (loudness) of each song as well as to set the max peak level.

BTW, I think the term "re-sample" has been confused with "word length reduction" (24-bit ---> 16-bit). Re-sampling is done only on files that are not 44.1kHz and that is the last step of the Mastering process (along with WRL and dither). It is a good practice to keep all files at the maximum possible sample freq and bit depth all the way from Tracking to Mastering until the final step going down to the Master part (CD-R for example).

It is also good practice to mix so that the peak levels are between -3 and -6 dBfs. As was mentioned before, if there is no headroom then the first thing the ME will have to do is knock the level down to a workable value. FYI - even applying subtractive EQing will cause peaks to rise. So if a mix already peaks at 0.0dBfs, then as soon as an EQ is inserted you get clipping So, a word of advice: If your tunes will go to Mastering, PLEASE don't Normalize!!! You're not gaining anything at all.

As for the quote from Mr. Katz:
That doesn't sound like something he would endorse (normalizing). I'll copy your quote and ask him what he means by it...

Hope all this helps...

[doulos]

i like going in by hand and pulling down all the peaks in a wav form to close match one another then run it through a multiband compressor attached to a spectrum analyzer bringing up the overall volume evenly to around +6 db. Good enough for demos, but if its serious i send it off for mastering and mix to -3 db and let them deal with it.

[Cloneboy Studio]

Personally normalization always seems to muddy up the sound... and I'm not quite sure WHY because it is just raising levels, not squishing transients or compressing... however, everytime I've tried using normalization it sounded "funny" to me.

I'll compress and hard limit myself.

Of course, I'm NOT a mastering engineer. My forte is mixing and tracking (in that order).

[mikemorgan]

[QUOTE=doulos]i like going in by hand and pulling down all the peaks in a wav form to close match one another then run it through a multiband compressor attached to a spectrum analyzer bringing up the overall volume evenly to around +6 db. Good enough for demos, but if its serious i send it off for mastering and mix to -3 db and let them deal with it.[/QUOT

I know it's a rudimental question, but how do you pull down individual peaks in a wav file? Can it be done in Sonar, or do you use another program? Can you literally "draw" audio to correct a momentary pop in a wave file? I need this one, I'm pulling my hair out on a demo project with just a couple instances of peak I really have to address.