There seems to be a different limiting effect from ISP limiting that feels more aggressive on the leading edge of a transient than doing just more regular limiting, but its hard to make this generalization across all tools and transient material.

A good limiter (and setting) keeps that leading edge lively and natural. This has been my struggle with ISP limiting, but I could also be losing track of how much extra limiting its doing.

Have you tried the isp on limitless also?

At least to me it sounds better than the ones I have tried before.. Turning it off is of course better sounding but probably worse sounding after conversion unless there’s a huge headroom. Haven’t found the right solution for me yet but as always it’s probably the balance.

Yes, defintely better than ProL2 and Ozone's ISP protection, but I can still hear it. Limitless sounds better without it IMO.

So what is the problem with a few ISP overs that are inaudible to a human if it means the track sounds better with its transients in tact?

That's exactly right. Since I master at 96khz and mostly output at -0.5dBfs without ISP limiting, I know there are going to be ISP lights going off downstream, along with some encoder error, but I really do not mind because the average ISP traveling over half a dB from a brick wall limit will register as less distracting to my ear on fewer transients than an ISP limiter will on more transients.

Are professional MEs typically mastering for Spotify? Like does Taylor Swift, just as an example, have separate Spotify masters?

I've yet to find a top charting commercial release that follows any type of loudness target.

Most MEs just make one master, and maybe deliver a hi res for Apple or less limited for vinyl.

No, as far as we know.

Excellent, that settles it for me.

I am not sure I am understanding all this correctly.

Let’s say I have a song the sounds great at 8.0 LUFS, integrated. I uploaded it to a streaming service which, on playback, pads it down (however it does that) to its “standard” 14.0 LUFS.

Now I take the same song and master it so that it sounds as best it can at 14.0 LUFS and then upload it to the same streaming service, which, presumably, plays it back (without padding) at 14.0.

Given the streaming is going to be MP3 (or some proprietary lossy method), will most people be able to tell that there’s a difference?

It's important to use the correct terminology to avoid speaking at cross purposes. The figure you're referring to is usually written as dB LUFS Integrated, and it's always a minus figure, e.g. -14 dB LUFS I.

Now that's out of the way, your two example files will sound similar in perceived loudness only, but the -14 LUFS I one will likely have 6dB more transient information, and I'd argue on a lot of material would be very audible (and better sounding) to many listeners, even with something like a crappy 128kbps stream.

[EDIT] A Picture speaks a thousand words... -14 LUFS I, then same file at -8 with loudness normalisation of -6dB engaged. See the lack of transients? You can hear it too, despite their level sounding very close.

Are you asking if one were to upload a -8 LUFS song and a -14 LUFS song to Spotify, would there be an audible difference?

If I'm understanding the question correctly, the answer is absolutely. The -14 LUFS song will be dramatically quieter in comparison. Just listen to two different songs that have different levels.

I just listened to CCR's Have You Even Seen The Rain (older) and then compared that to Father John Misty's Goodbye Mr. Blue (recently released). CCR clocks in around -13 LUFS and FJM around -9. Can you hear a difference? Of course. FJM is louder, Spotify's whatever-ever-they-do doesn't do too much.

My takeaway? Master louder than CCR if you want your music to sound comparable to modern music.

You say that your mix sounds great @ -8LUFS. Good. So that would be because the loudness you have chosen is the right loudness for the intent you have for the music. i.e the mix sounds best @ -8LUFS.
OK so level and loudness are not the same but completely inter-related in that you could just lower the level of the -8LUFS track to get to -14LUFS.
In that scenario while level matching there would be no important difference. You are just listening to the same track at different volume levels.

IF you actually re-master the track to -14LUFS then yes you will change the perceived tonal balance due to the wider dynamic range. Some sounds may appear to be higher in the mix and some may get lost. The higher level of transients may cause masking effects you don't like or like.
Preserving transients is not the be all end all in mastering. Getting them right is more important. Too lively transients can actually sound messy/distorted/slappy. Carefully sculpting them for the right amount of attack for the track is what you need to think about, not just preserving them.

Getting your track into its loudness sweet spot is the art. Mix master for the music. Make it sound great and forget about PLAYBACK NORMALISATION at the users end.

Hey Gregg, as somebody who dies a little inside every time I see a mixdown described as a "premaster," I certainly appreciate the effort to correct terminology. Can I offer just one small correction?

With LUFS, the dB is unnecessary, as the LU replaces what we used to refer to as dB RMS.