Poor Mastering?
Jan 23, 2016 at 1:33 AM Post #31 of 34
Rick Rubin is a tard.
 
<close thread>
 
Jan 23, 2016 at 2:29 AM Post #32 of 34
   
No, that's not really the workflow. The band/producer does not send the mastering engineer a rough mix, they send a completed/final mix. The mastering engineer effectively optimises that mix for the target consumers. That (in theory) is the end of the processing, there should be no other processing once the mastering engineer has finished with it, no other "post production unit" which makes any changes at all. This isn't necessarily true with vinyl production though, where some processing (say the RIAA curve) may be applied after the mastering engineer has finished.

To add to this, on many productions the mastering engineer is making very little if any adjustments to CD quality and above. The mastering engineer optimizes the recording for each format it is to be released on. In the 80's and 90's you would go to the mastering studio with final mixes you would do the CD master first, if needed they might make some minor eq and and dynamic adjustments, maybe clean up  the start or end of song. They might sequence the song order and adjust the timing between songs.
 
That was your best highest resolution master. Back then you also needed a cassette master, they would maybe roll off the low end, eq some to compensate for the limitations of high speed duplication. You might have even needed masters for cassette singles. Then you need to cut vinyl. First you need an LP master, they will roll off the low end sum the low end to mono, if you had phase issues they will attempt to correct them as best they can or it will cause a ton of problems cutting the vinyl. If the recording engineers carefully checked phase by ear, meter or oscilloscope there should not be any problems.
 
They cut the LP and some test acetates the quality is the next step up from cassette. You might also have 12" 45 singles or eq's you can cut deeper and wider and have plenty of space between the grooves this is your highest quality vinyl. Then you cut the 7" singles these can be good quality as well, if the song is short, as the song gets longer you have cut the grooves closer reducing the quality. You could leave the mastering studio with dozens of masters each slightly different in quality but in theory each maximized in quality for each format.
 
As cassettes and vinyl went away in around 2000 till now they just made CD masters. When Apple came out with mastered for iTunes it confused many people, because it was forgotten even by industry people that the mastering engineer attempts to maximize the quality for each format. Apple gave mastering engineers a tool to check their masters for encoding quality and possible encoding errors. The iTunes master could be exactly the same as the CD master or they might have made some minor adjustment to optimize the iTunes master. 
 
Today you might have a CD,  "high rez", lp, 12', 7" iTunes, MP3, ringtones, and streaming masters. I have heard of recent releases having over 500 separate masters.  
 
Jan 23, 2016 at 4:25 AM Post #33 of 34
  Just to drive home my point that dynamic compression is a sliding scale, which, somewhere on the extreme end becomes synonymous with clipping, I prepared a few more graphs.

 
I think you're missing the point that no one in their right mind would set a limiter's peak output to 0dB, they are always set to some point below 0dB, even if it is only -0.1dB. The output of a limiter may look very similar to a clipped signal but in most any DAW we are not looking at an audio signal we are looking at a graphical representation of digital data, which merely resembles an audio signal. In practice, a limiter obviously distorts an audio signal but does not induce digital overload distortion, unless the engineer is not using an oversampling limiter, in which case overload distortion is a possibility due to inter-sample peaking. In practice, limiters do not suddenly truncate from one sample to the next, as occurs with clipping, they compress more gradually towards the peak output setting, even though it can sometimes be hard to tell the difference on a graphical representation of digital data.
 
Quote:
  Actually there was a (supposedly actually true) communication with the mastering engineer, who said in no uncertain terms that the mix was brickwalled before it got to him and that he would never have put something this bad out, so again we're really not blaming the mastering engineer.... But again this isn't a genre issue, because even metal-heads were complaining about this one, especially once Guitar Hero gave them a chance to compare to a different mastering. Listening to the rest of Metallica's oeuvre, I don't hear a precedent for "hey, lets have the album clip literally every time the drums are played."

 
That seems to clear-up that in this particular instance we are talking about clipping rather than limiting. I haven't listened to the album in any detail and in the original posting of the waveform, the track/album was not given. Nevertheless, what you say still raises a couple of points:
 
1. Broadcast TV audio is always brickwalled, although generally less severely now than it was a couple of years ago. Some of the waveforms didn't look vastly different from the waveform posted in this thread and yet TV audio is generally not terrible ... and when it is, it's generally due to some factor other than the brickwalling. Standard TV audio specs used to require mixes be brickwalled at -9dBFS, so while we had brickwall artefacts there was never any clipping (digital overload) artefacts. So when the mastering engineer stated that the mix was already brickwalled when he got it, what exactly does that mean? It could have been brickwall limited but not clipped, in which case the mastering engineer has caused the clipping. I take what he said to mean that the mixes were clipped when he received them (rather than just brickwalled), in which case it's not his fault.
 
2. While you personally are "not blaming the mastering engineer", it seems clear that limpidglitch was. Even accepting that the waveform is clipped rather than brickwall-limited, it's still not certain that the mastering job was poorly executed. We don't know how badly the mix was clipped, maybe the mastering process reduced the clipping (relative to the mix) and what remains could not be fixed any better than he accomplished and therefore the mastering engineer actually did a very good job. Maybe a re-mix was done (with less/no clipping) which was not available at the time of the original mastering. There are various possibilities/scenarios which mean we cannot just blindly blame the mastering (or mastering engineer), regardless of the fact that poor mastering is a distinct possibility or even a probability.
 
G
 
Jan 23, 2016 at 5:10 AM Post #34 of 34
 
I think you're missing the point that no one in their right mind would set a limiter's peak output to 0dB, they are always set to some point below 0dB, even if it is only -0.1dB. The output of a limiter may look very similar to a clipped signal but in most any DAW we are not looking at an audio signal we are looking at a graphical representation of digital data, which merely resembles an audio signal. In practice, a limiter obviously distorts an audio signal but does not induce digital overload distortion, unless the engineer is not using an oversampling limiter, in which case overload distortion is a possibility due to inter-sample peaking. In practice, limiters do not suddenly truncate from one sample to the next, as occurs with clipping, they compress more gradually towards the peak output setting, even though it can sometimes be hard to tell the difference on a graphical representation of digital data.
 

 
If you look at the code and the plot of the transfer function you'll see that I set the limit at -6dB.
Yes, normally limiters are used more sensibly than this, but this is what you have to do to make a limiter produce the flat tops and the ringing seen in the various plots I've posted.
 
 
2. While you personally are "not blaming the mastering engineer", it seems clear that limpidglitch was. Even accepting that the waveform is clipped rather than brickwall-limited, it's still not certain that the mastering job was poorly executed. We don't know how badly the mix was clipped, maybe the mastering process reduced the clipping (relative to the mix) and what remains could not be fixed any better than he accomplished and therefore the mastering engineer actually did a very good job. Maybe a re-mix was done (with less/no clipping) which was not available at the time of the original mastering. There are various possibilities/scenarios which mean we cannot just blindly blame the mastering (or mastering engineer), regardless of the fact that poor mastering is a distinct possibility or even a probability.
 
G

 
If the damage was done in the mix, then the mastering engineer has been seriously over-paid, because that waveform has not been touched after it was limited/clipped.

 

Users who are viewing this thread

Back
Top