Loudness Wars

Moderators – This was my best guess for where this thread belongs, but feel free to move it to a better place.

I have been following some recent posts on Darko Audio on the subject of source material dynamic range. The included links cover the discussion. I may not agree with everything he says, but it has gotten me thinking. What do you guys think of dynamic range vs loudness? Does this explain some of our personal preferences for detail vs warmth? Do you think it would extend beyond source material to equipment choices?

https://darko.audio/2023/08/when-hi-res-audio-stops-making-sense/

https://darko.audio/2023/08/when-hi-res-audio-stops-making-sense-a-follow-up/

https://darko.audio/2023/08/when-hi-res-audio-cant-compete-with-vinyl/

As for me, I looked up some of my favorite albums/songs on the dr.loudness-war.info database. I only wish the database had more entries. I found the majority of my favorites were in the green or having high dynamic range. That probably fits with my NOS Pavane dac preference and my leaning toward small group acoustic players.

I also wonder if DROffline Mk II could be used to measure some differences we currently aren't able to explain with other measurements. It strikes me that most of our measurements are made at a single frequency or at uniform loudness. For example, music or a non-uniform loudness test signal being played through a test subject dac, then fed into an ADC and saved to a computer for DROffline measurements. That dac could then be compared another dac with this measurement score.

 

I believe the pursuit for a high DR number in roon is a wrong way to approach a master. Instead they can feel different and have a separate intention for the lister.

"Lower" DR, say 7 is a nice sweet spot, stuff past 12 is very absurd and similarly to something below 4. While the background and feeling of sounds popping out of the sound scape is good. it feels unnatural to my ears but I understand the desire of a distant sounding vocal chain and small things like coughs appealing to many audiophiles.

Gotta ask what is your intention when listening?
Leaning in or leaning back?

 

I don't use Roon, so forgive my ignorance. Does Roon report or let you search on a DR number? That's interesting.

I agree that saying a high DR number is both not for everyone. I do think that DR is naturally going to vary with the type of music, so I couldn't target just one DR level, but might develop a preference DR number for different types of music if I had access to DR numbers. I did like Darko's example of 4 different masters of one recording and how the DR number of those recordings correlated with his ranking of the masters.

And to answer your question. Leaning in or back depends on the music and what else is happening. But more often than not it's leaning back.

 

report, yes. I don't think you can search based on DR though
-----edit------

Interestingly enough, I was just paying attention to this last night. Listened to Best of Santana (16/44/1 DR5) and then roon radio played one of the same songs from Santana's Greatest Hits album directly afterwards (16/44.1 DR5).

I was stunned how different they sounded and the Greatest Hits version was much quieter across the board. I assumed the Greatest Hits version was an older mastering that hadn't been compressed vs the newer Best of version. But both albums shared the same DR and looking at the info Roon reports for the specific song that came on:

Jingo - Greatest Hits (1998) Track Gain -7.3dB Album Gain -9.1dB DR128 3.9
Jingo - Best of (1992) Track Gain -13.0dB Album Gain -14.9dB DR128 3.7

Not sure what the above means, maybe the more knowledgeable here can figure it out, but it did peak my interest to go digging...

 

It is a common misconception that "loudness wars" is a thing that was born in the digital age. It isn't. It was born long before that. 1950s? Maybe even earlier. Radio stations wanted to catch and keep hold of passing ears that were spinning the tuning dial. It has been known about, and complained about, since then.

That's a strange thing to say. Compressed, in the dynamic-range loudness sense, is a distortion of the music. The natural music is not for everyone?

I'm sure there is uncompressed music that does not a very high dynamic range. Some music has lots of loud/soft in it; some music not so much. I doubt that anyone could chose their music by DR number.

I am not talking in absolute terms, especially as I am not a recording engineer. I'm sure there are times and tunes where dynamic compression is actually necessary. I' guessing that there may be some passages in classical music that we would not even hear on a recording without it. (and others that would blast our speakers to bits)

However, I'll stick my neck out and say that, talking in layman-listener terms, there is absolutely no reason nor excuse for compressed low-DR music, and that no audiophile worth the word would prefer it, unless their way of taking time off from audiophilia is intoxicated head-banging sessions!

I thought this was all done and decided (by music lovers, but sadly not by music companies) in 1970-something.

 

I came across a fascinating talk and demos by producer and mastering engineer Alan Silverman, about dynamic compression, loudness wars, and how streaming has upended how music is being mixed.

 

Great video! This is exactly what I’ve been thinking for some time now: Compressed music doesn’t make sense anymore in the era of normalized streaming. And I do think that today’s masters are much better than those from the nineties. I hope they will become even better as this knowledge trickles down to every audio engineer.

 

Good compression and peak limiting are awesome.

“Dynamic range is the enemy.” - Geoff Daking.

If the listener has to change the volume, the mixer or the masterer have failed! You don’t have to change the volume at an acoustic concert!

Unfortunately what matters is how you do it and how much of it you do. Slamming up against digital zero is not the same as good compression. There are Motown records that are stupid loud and distorted. That Red Hot Chili Peppers record might be slammed and have some peak clipping but surely is going to be very hifi compared to Californication, Taylor Swift, and maybe not as fuzzy and lame as lot of Motown masters where the album masters were often wimpy but the singles were cool, distorted, and slamming.

Traditional compressors and limiters are very smoothed over volume modulators. Think a synthesizer effect but much cleaner; the better ones hug the material very well when properly used. They turn down the volume of the entire sound by the ratio when the sound is above threshold and then turn it back up again when the sound is below the threshold or changes. The attack is the speed at which they turn down the volume, the release the speed at which they turn it back up, and they constantly switches between attack and release based on whether the signal is rising or falling. Attack and release are usually one very complex low pass filter with various slopes applied to the control voltage to slow down the gain reduction. Think about it. It works. Analog limiters are just very high ratio, very fast attack compressors (think an 1176) or clippers with gainstaging that accounts for the Gibbs phenomenon to not overload something later on.

Digital limiters are much more complex and usually much more distorted because of aliasing and trying to do a whole lot more than an an 1176 right after a mic pre or a Fairchild 670 or TAB u73b in old vinyl mastering setups.

These digital limiters aren’t just turning down the peaks; they are leveler compressors, limiters, and clippers all merged into one mega distorted plugin. They have lookaheads that modulate the volume based off what happens in the future rather than just what happened in the past like normal compressors. This sounds great in theory but how it must be accomplished to not release prematurely is by delaying the control path and making the attack hold without releasing for the length of the delay. If you think this will sound like crap on a snare rimshot, you’re right because the gain reduction will look like this “\_/” instead of a V with a steeper \. Modern lookaheads involve using the delay to take an average or a maximum peak value and scaling the entire attack to reach the value in a minimum amount of time equal to the delay. So instead of \_/ they have an imbalanced V where the \ is the length of \_ instead of a steeper \. That’s still more potentially unwanted volume modulation than a compressor that can switch instantly between attack and release based on the polarity of the signal.

Take why Alan Meyerson used in the video:

PSP Xenon. This is a standard leveler and a very low distortion lookahead limiter that he didn’t set for low distortion. He didn’t enable the higher quality settings at the top left, chose mode A which is a straight pcm sample clipper, instead of mode B which tries to emulate an old Waves digital limiter or C which is PSP’s own pretty clean algorithm and didn’t use the leveler which is slightly distorted but does the job fairly well enough.

Fabfilter Pro L2
This is a leveler, limiter, and clipper. It is very colored as a whole process and can be set to be rather low distortion and not that clippy but who knows?

DMG Limitless.
This is a multiband limiter with a clean RMS leveler, a very dirty peak limiter, and a clipper that’s just okay.

obviously you can set them to be cleaner or more distorted, not use all of the functions, or use something else that could be perceptibly much cleaner like use a nice mastering compressor and clip off any extraneous annoying snare peaks or screams you’re not afraid of fuzzing out just to not modulate the rest of the audio where they cause introduce gain reduction. The nice, well controlled mastering compressor will have cleaner overall action and the clean or cool sounding clipper will have more distortion but less modulation during very brief peak events than a lookahead limiter. Or you could replace the clipper with an an ultra fast limiter that makes the modulation time incredibly brief but these are rare in both digital and analog.

Of course that isn’t as tidy for a short YouTube video and won’t get you a solid red brick wall on a computer screen, which is what most of the ignorant musicians, producers, and record labels want. They don’t want something that sounds ideal; they want something that looks like everyone else. They want to conform.

 

What is up with that album? Californication has great songs, but it is the only album I have that has such serious audible clipping. Were they all really that high when they mastered it?

 

Perhaps the better question would be how high were they...