Dynamic range DR measurement and loudness war

[Rockerilla77]

Hello to everybody,

googling about "DR measurement + loudness war", I found this forum. I'm so impressed by the discussions on this forum. I'd like to share this:

NONPARAMETRIC ESTIMATION OF THE DYNAMIC RANGE OF MUSIC SIGNALS
by Pietro Coretto and Francesco Giordano

That one is the arXiv preprint, but it's almost the same as the final version published here.

 

[frabor]

Nice, great paper, tks.

In case you are not familiar, take a look to the dynamic range database

 

[Rockerilla77]

[...] In case you are not familiar, take a look to the dynamic range database

Thanks, I know it. The paper shows that the DR statistic (TT Dynamic Range Meter) sometimes have strange behavior (see the examples of Figure 8 in the paper). The only problem with this kind of measurement is that it is computationally demanding. Certainly, it won't be feasible for home/non-power users.

The project is to develop a simpler version of the estimating algorithm that is feasible for fast software implementation. Actually, the idea was to also work out something that looks at the spectral distribution of the compression. From what I know, there is no literature on this.

... the only problem is the lack of time!

 

[frabor]

As it always is

 

[Cbdb2]

I don't agree with this statement: "Loss of DR along the recording-to-playback chain translates into a loss of audio fidelity." Compression is a necessary part of most recording, and used properly increases fidelity.

Measuring the dynamic range of a electronic signal is not the same as measuring the dynamic range a person hears. Time constants (which they don't consider) matter, is it a limiter or a compressor, very different sounding. You won't hear limiting on very fast transients ( energy is power x time). Frequency content matters. Compress the low end and some music will sound more dynamic.

 

[DVDdoug]

I don't agree with this statement: "Loss of DR along the recording-to-playback chain translates into a loss of audio fidelity."

How about this? "Loss of DR along the recording-to-playback chain can, and often does translates into a loss of audio fidelity quality."
It's a matter of taste and lot of people are disappointed with Loudness War remasters (and other "over compressed" releases).

Compression is a necessary part of most recording, and used properly increases fidelity.

Nothing is "necessary". A lot of live music is performed without compression. But some compression/limiting seems to be preferred.
Technically, it lowers "fidelity" if you define fidelity as being faithful to the original sound. ...But with modern studio recordings it's kind-of hard to define "original".

Measuring the dynamic range of a electronic signal is not the same as measuring the dynamic range a person hears. Time constants (which they don't consider) matter, is it a limiter or a compressor, very different sounding. You won't hear limiting on very fast transients ( energy is power x time). Frequency content matters. Compress the low end and some music will sound more dynamic.

Somewhat true. There is no one-way of defining or measuring dynamic range but that doesn't make the measurements totally useless. "DR" is often defined as the crest factor and there are flaws with how it correlates to perception, but it still isn't necessarily useless. EBU R128 "Loudness Range" is probably a better (but still imperfect) measurement because it tries to compare the variations is perceived loudness.

 

[frabor]

I do agree that compression, like autotune, is a method for producers to shape the recorded sound. Quality is a debatable aspect and fidelity is the process to capture the signal as close as the listener would hear it on the optimal location.

Rock and EDM used compression on all of their recordings; EDM often uses low DR and high compression; classical music uses low compression and high DR. Vocal pop uses autotune and many tricks to create a new expanded version of the vocal track.

Neither autotune or compression increases fidelity, since they create a new version of the track. Both are tools ,among many others, for the producers to manufacture a sonic signature. The fidelity cames from packing the multitrack master into a similar experience in a stereo track; a high fidelity transfer from the recording master.