Mastering close to 0dB - is it bad?

Hi guys,

 

I have 2 masters of the same audio - 1 was mastered with -0.4db True Peak/ISP & 1 with -0.1dB True Peak/ISP.

 

They sound slightly different, the one with -0.1 dB has a little bit of 'oomph' to it.

 

Strangely enough, they both have the same integrated loudness, they both are -13.5 LUFS.

 

Through iTunes, I converted both audios to the worst possible mp3 option I could make (125kbs) - I wanted to see how they perform at their worst - and auditioned them.

 

The one with -0.4dB now peaks at 0dB. The one with -0.1dB now peaks at 0.5dB.

 

Listening to both mp3s closely, I'm liking the one that peaks at 0.5dB better.

 

Question is, do I trust my ears? Is little distortion can be pleasing?

 

Because they both are -13.5 LUFS and volume matched but I seem to favor the one that has a subtle distortion to it.

 

Another question is, what advantage does the one with -0.4dB True Peak/ISP have over the one with -0.1dB True Peak/ISP?

 

Thanks in advance and happy holidays.

I have 2 masters of the same audio - 1 was mastered with -0.4db True Peak/ISP & 1 with -0.1dB True Peak/ISP.

 

They sound slightly different, the one with -0.1 dB has a little bit of 'oomph' to it.

The extra "oomph" might simply be the extra perceived loudness or peak energy. A 0.3 dB difference can be discernible in an AB test, but have you tried performing a real blind test to eliminate confirmation bias?

 

Strangely enough, they both have the same integrated loudness, they both are -13.5 LUFS.

Loudness meters generally have a precision of ±0.2 LU and a small peak difference like this could be offset by a combination of other things that happening during limiting and the aforementioned precision tolerance.

 

Through iTunes, I converted both audios to the worst possible mp3 option I could make (125kbs) - I wanted to see how they perform at their worst - and auditioned them.

 

The one with -0.4dB now peaks at 0dB. The one with -0.1dB now peaks at 0.5dB.

I believe that's 128 kbps, not 125 kbps.

 

To avoid lossy conversion overs it's generally recommend to set a ceiling of -1 dB True Peak. However, the best way is to perform an offline check (Apple's realtime AU RoundTripAAC isn't precise enough and will vary its readings), this is done via the Terminal and the "afclip" script, which is part of the MFiT download package from Apple.

 

Listening to both mp3s closely, I'm liking the one that peaks at 0.5dB better.

The same one you liked from the beginning so that's not surprising even though you're now getting a bit of distortion. Depending on the song it could be virtually inaudible so all you're experiencing is a bit of extra perceived loudness or peak energy.

 

Question is, do I trust my ears? Is little distortion can be pleasing?

It can. Usually pleasant distortion is associated with even order harmonics or at least a blend between even and odd. But this kind of clip distortion is a bit unpredictable as it depends on the headroom of the DAC playing, so in this case it's not as simple as trusting your ears, at least from a theoretical standpoint.

 

Another question is, what advantage does the one with -0.4dB True Peak/ISP have over the one with -0.1dB True Peak/ISP?

Less lossy conversion clipping. This clipping and/or inter-sample peaks could lead to some algorithms behaving less predictably when loudness normalizing, but in my experience it's nothing to worry about in real life, e.g. Spotify or YouTube.

Thanks for the feedback lagerfeldt.

 

The 1 that peaks at -0.4dB True Peak/ISP was done by an online friend of mine and the 1 that peaks at -0.1dB True Peak/ISP was done by a rather well known online mastering studio.

 

They were told to do minimal/light compression and target the release for online services (spotify, youtube, etc etc).

 

To avoid lossy conversion overs it's generally recommend to set a ceiling of -1 dB True Peak

 

Question is, since I'm liking the sound of the -0.1dB better, do I tell the mastering engineer to lower it the generally recommended ceiling?

 

I don't want to sound like i know better than the mastering engineer and i have the feeling (from the higher fees, his 10 years plus of experiences) he might laugh at me for pretending to know things better than him?

 

Should i just accept the master that I like or should I tell him to revise it to follow generally recommended ceiling of -1dB True Peak?

 

Thanks in advance.

Thanks for the feedback lagerfeldt.

 

The 1 that peaks at -0.4dB True Peak/ISP was done by an online friend of mine and the 1 that peaks at -0.1dB True Peak/ISP was done by a rather well known online mastering studio.

I was under the impression that we were talking about the same mastered version and same limiter settings, only with two different peak levels.

 

Since that's not the case then all bets are off, i.e. the minor peak difference here is completely negligible compared to all the other processing differences. No wonder you prefer the professionally mastered version.

 

You should choose the version that sounds the best to you.

 

They were told to do minimal/light compression and target the release for online services (spotify, youtube, etc etc).

Since Spotify and YouTube use two different loudness targets that's a bit tricky, but I would simply aim for something that sounds good. Currently a target somewhere around -11 LUFS (I) and a peak level of -1 dBTP should work fine on both platforms, although Spotify now aims for the slightly lower -14 LUFS (I) target.

 

Spotify used to be -11 LUFS (I) but now uses an absolute target of -14 LUFS (I) like all other major streaming platforms. The exception is YouTube and they've made several changes to their target, but currently it's somewhere around -11 LUFS (I) to -12 LUFS (I).

 

YouTube will currently not raise the level of your audio if it's lower than the target, but Spotify will raise the level with a limiter which is why you definitely want to control that yourself.

 

To avoid lossy conversion overs it's generally recommend to set a ceiling of -1 dB True Peak

Question is, since I'm liking the sound of the -0.1dB better, do I tell the mastering engineer to lower it the generally recommended ceiling?

 

I don't want to sound like i know better than the mastering engineer and i have the feeling (from the higher fees, his 10 years plus of experiences) he might laugh at me for pretending to know things better than him?

 

Should i just accept the master that I like or should I tell him to revise it to follow generally recommended ceiling of -1dB True Peak?

A ceiling of -0.1 dBTP is fine as well. The difference will be minimal. I've done thousands of masters on Spotify that originally peaked at e.g. -0.1 dBFS (that's FS, not TP) and they sound fine.

 

That being said, best practice in 2018 would be to use a ceiling of -1 dBTP for streaming services to avoid lossy encoding overs.

Thanks for your reply lagerfeldt.

 

Since you've done thousands of mastering works, I shall take your advice - accept the master that I like and move on.

 

I think the reason why both engineers didn't hit 11 LUFS/-1dBTP is because the genre of the track - it's orchestral/classical+some modern twist to it. The track has to remain dynamic.

 

Thanks again lagerfeldt, you saved me from embarrassing myself to ask for a revision, lol.

Ah yes, well then a target of -11 LUFS (I) could be way too high. Let me know when the track's out on Spotify!

Great thread with some great opnions / tips. 1 thing I don't understand: if True Peak is to make sure that intersample peaks that may appear while DA converting are taken into account, why shouldn't we use a True Peak of 0dB - isn't it a bit of repetitive to use True Peak ánd a safety margin (-1dBTP)

Apart from intersample peaks you also have lossy encoding overs to deal with, so I suppose the idea behind a ceiling of -1 dBTP (which is also the standard for R128 material) is like a safety-safety margin. In my experience that's overkill, but there you go.

 

Since it's been some time since my original posts in this thread I should update some of the numbers I've referenced:

 

· Spotify is currently using ReplayGain + a 3 dB offset that aligns with approximately -14 LUFS (I). This is true for Spotify Free and Spotify Premium with the normal/default loudness normalization setting. However, Spotify Premium with the loud setting uses approx. -11 LUFS (I) and Spotify Premium with the Quiet setting uses approx. -23 LUFS (I). Spotify has informed that they'll move to actual ITU-BS.1770/LUFS sometime in the future.

 

· YouTube is around -13 LUFS (I)

 

Going directly for these targets is often a bad idea despite what a lot of well-meaning people claim.

 

In most cases you're well off with a pop/club/urban/rock master around -11 LUFS (I) to -9 LUFS (I) and keeping inter-sample peaks below 2 dB. If you're doing more dynamic music and aiming directly for -14 LUFS (I) then be aware that Spotify has two limiters: the first limiter raises the level of any masters below -14 LUFS (I), so make sure you hit just above that target to absolutely avoid the limiter. However, you'll still get limiting with the Spotify Premium loud setting since you didn't hit -11 LUFS (I). Then there's a second limiter which acts as a protection limiter. The second limiter kicks in on levels above -1 dBFS post encoding. The "post encoding" part is important because this limiter is unlikely to affect most -11 LUFS (I) to -9 LUFS (I) masters because they've been lowered by several dB below this ceiling, even with some ISP overs.