the shakira example is a classic case of what the codec was never designed for. mp3 uses a psychoacoustic model that discards information it thinks you won’t hear, but when the source is already pushed hard and full of transients, the encoder is making bad decisions on top of a signal that has no headroom left for them.
the intersample peak problem is the more insidious one: your master might show -0.1 dBFS on the meter, but the analog reconstruction after D/A conversion can spike well past 0 dBFS depending on the waveform shape between samples. most streaming encodes (aac especially) make this worse. targeting -1 dBFS true peak instead of -0.1 dBFS is a small concession that saves you from audible distortion on a lot of consumer playback chains.
I was taught by a dolby sound engineer that your mix should be good enough that a mastering engineer doesnt have to do anything other than set a limiter for the desired loudness. Thats it.
As a mastering engineer for the last 20+ years, that’s exactly how I approach mastering my own music too. A bit of Limitless to get the volume where it needs to be, anything else I can adjust better in the mix down.
I think it’s really good to highlight it because there is this sense that you should take your mix so far and then bounce it to stereo and then start EQing it and compressing it etc, but it has never made sense to me.
If the bass is too loud rather than try and EQ the master I will just go to the project and pull down the fader on the bass bus.
I try to get there with my mixes and when possible I ask clients for mix revisions as well. Sometimes that can happen and I think the master always turns out better because of it.
Sometimes it can’t and then I do my best with the material and my bag of tricks.
You joke but that really is the ideal. I don’t do a lot of mastering these days, I actually see myself more in a “technical producer” role. Like instead of walking into the studio in the middle of the day and saying " this needs bongos" before disappearing for the next 3 weeks, with the best artists I work with I get paid to basically say “low shelf out a db on the pads under 130hz” and then I put 3 band eq and a limiter on it.
There are people who don’t get that close and I do more work for them. But I always give them the feedback and try to get them closer to that ideal. I’ve noticed that after 2-3 albums most of them get there and I don’t have a whole lot of notes anymore.
