What Does a Mastering Engineer Require?

[masteringhouse]

yeah, that's what I do.

My question that I guess I did not articulate well enough is this : what is the optimum level for introducing the least amount of digital gain/reduction artifacts? At what level is a track in unity with the actual recorded signal? Or does that even matter? Obviously, to get the tracks as pure as possible, one would theoretically keep as close to this as possible, hence, the highest pre-clip level possible (full mix)... right...?

Does this make sense?

1 bit represents approximately 6 db. So if the maximum level you recorded at was -6, you would only be using 23 of the 24 bits (or 15 of the 16 bits) i.e. not full resolution. Of course less than -6 brings your resolution down further.

In order to get the best resolution (which is what I think you are getting at) you would try to get as close to 0 dBfs as possible without actually getting there. Remember a digital over is not really an "over". It is a measure of how many consecutive readings you have at 0 dBfs. This can vary from one digital device to another, so to play it safe I would only go to -1 dBfs maximum to ensure that you aren't clipping.

 

[greggybud]

lotuscent -


Another point about bus processing (both EQ and compression/limiting). When using processing on the main stereo bus that is going to be applied in mastering, you are causing the audio to go through the same calculations more than once. This introduces quantization error distortion (see my previous post).

This is one of the reasons why normalization should not be done in the pre-mastering stage (if it should be done at all). You are going through a series of calulations to bring the level up to 0 dBfs when the overall volume is going to go through another series of volume calculations when being compressed, EQed, etc. and is going to be brought up again. This makes digital audio start to sound grainy and thin. Likewise, bringing the level up via a compressor/limiter to a point that isn't going to be used in the final product is a waste.

I thought normalization was a nasty no no, but then read where it was okay to normalize IF you are at 32 bit floating point. I have the article somewhere and I cant explain it as it was extremely technical. Does anyone have an explanation?

 

[sonicpaint]

Try not to think of the level so much.

I agree with everything that masteringhouse said. In short the best thing anyone can do is just mixdown to what ever format is being used without doing anything but that. Keep your processing to the minimal and if you can, avoid it all together.

Here is a good example of what to do. I'm going to refer to Digital mixing here but you'll get the idea.

Listen to your song and mix it to taste using all the tools that you normally use during tracking. Make sure that the song isn't clipping anywhere. Then save those setting and create your mixdown. That's it!

Notice that I didn't make any other changes while the mixdown was happening? That's what's trying to be explained. Don't do anything to it but just create one file (song) from your tracks keeping a close eye on the meters, for an error free mixdown. I know what you meant by the level thing during mixdown and I'll try to help you understand it's not a big issue in most if not all cases.

The normal dynamics of your any given tracks in a sessions song will very. Though this is the case, the overall dynamics(in most cases) will not be so drastic as to cause your mixdown to be at a level where the signal to peak ratio is on the least favorable(meaning more noise due to small signal to peak ratio). The only time that this is an issue is when you have a peak that for what ever reason (errors, clicks,pops) are above the average by 8db or more. If you find that this is the case then find that peak or peaks in those tracks (no matter how long it takes) and fix it before your mixdown.

So don't worry about the level so much as to the quality that you'll be handing in to the Mastering Engineer. You'll be and the engineer will be really happy you did.

I hope this time we hit the nail on the head for you lotuscent.

Good-luck
sonicpaint

 

[jake-owa]

1. Don't fade in or out
2. Add NO compression or EQ to the mixes
3. Save multiple versions of each mix to chose from, at least include a vocal up mix in addition to the regular mix
4. Bring the ME data and not audio versions of your mixes
5. Label everything meticulously
6. Make a list of the desired edits and sequence of tracks
7. Have a good long talk with the guy and let him know what your favorite albums are, what you like or dislike about music and do not go away unsatisfied

Oh yeah, I forgot, when mixing to analog leave a good space between songs and include test tones on the tape.