In my recent test on the Spotify loudness normalization I was (once again) reminded about that there is a limiter in their signal chain.
The "word on the street" has always been that it was set at -0.1 dBFS or 0 dB but after I discovered that Spotify had dropped their loudness target by -3 dB in May (2017) and went on to do some measurements and test... then I did see some evidence that suggested that the ceiling was not set at -0.1 dBFS (and that it WASN'T really good at limiting either). But I didn't think much about it until I did some test last week when we (me and Ian Stewart) did some test on ReplayGain and Spotify loudness management. It became even more clear that the ceiling wasn't -0.1 dBFS (and it probably is responsible for some of the "margin of error" we got on material that had to be turned up).
So after sending an inquiry to Spotify, they replied and offered this information:
The limiter's ceiling used to be set at -0.04 dBFS, but we have now changed it to -1.0 dBFS. That change was made in May when they lowered the loudness level.
(Additional info on the limiter settings: Attack 5 ms, release 100 ms).
We have discussed True Peak limiting, and it is obvious that it's needed. We should measure True Peak levels, and we should make the limiter True Peak sensitive.
Both of these points are unfortunately not high on the "To-do-list", so I don't think it will happen all too soon.
Now, so far, we have decided to go with the limiter - but we think it's more important that we balance the volume / loudness with no limiter... especially as we've lowered the level so fewer songs will be limited.
You can turn off the loudness adjustment completely, if you really want to get rid of the limiter.
So, yes... for time being at least, then there is a limiter in the chain but it's actually very rare that it's working. On the material that I know for sure it has affected then I didn't hear any negative affects of it, but like with all audio processing ... results are program dependent ;)
That's it.... I hope this helps.