Skip to main content
Topic: Questions on "loudness wars" and/or bad CD engineering - doing a FLAClibrary (Read 927 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Questions on "loudness wars" and/or bad CD engineering - doing a FLAClibrary

tiny bit of background: I used to listen to MP3's for convenience, but rejected them, and listening to music entirely due to wearying artifacts

I recently used MediaMonkey & Mp3Tag to re-rip all my CDs to FLAC, and I'm still hearing TWO (what I perceive to be) different artifacts (in some albums) that I am interpreting as having been introduced in the engineering phase.  One is general "noisiness" -- the worst offender I've found being "The Killers" (any album!), and the other seems to be a selective attenuation of A TRACK in the mix in order to "make room" for another -- usually a bass track.

I have been looking for an audio analyzer that would show music with the x axis being time (as it goes by), the y axis being frequency (high at the front to low at the back) and z being amplitude or volume.  I think this would show a held chord or note being dropped a split second BEFORE a bass note hits -- the sort of thing that might be done with a cheap DSP in a purpose-build "Arduino" sort of product.

I've read the term "pump" being used here and elsewhere and it certainly describes the gawd-awful artifact that I find so irritating, but I don't know enough of the vocabulary to identify it in a way others readily understand.

Is there a primer on the common vocabulary used for this stuff?
Is there a free or affordable audio analyzer that would use the x-y-z axis presentation I described?
Are these artifacts real, or are they in my head?  :)
Thanks for your patience!

Re: Questions on "loudness wars" and/or bad CD engineering - doing a FLAClibrary

Reply #1
Audacity (a FREE audio editor) has an optional  Spectrogram View.

That may, or may not, be helpful.   It's usually easier to hear something in an audio recording than to see it.    And if you haven't done so already, you may be able to zoom-in and see the defect in the regular waveform view.  

...And then the issue is, whadda' you gonna' do if you see it?

Quote
I recently used MediaMonkey & Mp3Tag to re-rip all my CDs to FLAC, and I'm still hearing TWO (what I perceive to be) different artifacts (in some albums) that I am interpreting as having been introduced in the engineering phase.
Lots of things can go wrong during the production process.     Whatever you're hearing, the producer/engineer may not have considered a defect, or maybe it wasn't worthwhile (or possible) to fix or re-record, etc.

...Every time I've though I heard an MP3 compression artifact, it's turned-out that the CD had the same "defect".

Quote
I've read the term "pump" being used here and elsewhere and it certainly describes the gawd-awful artifact that I find so irritating,
Dynamic compression (not to be confused with file  compression like MP3) is a kind of fast automatic volume control...   If something loud comes-along (like a loud kick-drum) the volume is momentarily turned-down and sometimes you can hear the volume "pumping" up and down.    Or, there is something called "ducking" where the bass guitar is turned-down momentarily with every drum kick, etc.    If it's done right you shouldn't hear the side-effects, but it can be done wrong or over-done.

Quote
I think this would show a held chord or note being dropped a split second BEFORE a bass note hits
There is actually something called "look ahead", and if done right it can allow compression/limiting without distortion and perhaps with less pumping).   That is a feature of digital processing...  The older analog compressors didn't have memory or delay so they couldn't "look ahead" and they can only react to what's happening or what has already happened.

Re: Questions on "loudness wars" and/or bad CD engineering - doing a FLAClibrary

Reply #2
Another source for general noisiness in loud albums is Clipping. It sounds like a few clicks during loudest monents, identifyable especially on bass or vocals, or a sustained crackle. Spectrally complex sections with multiple instruments sound like a fuzzy noise was mixed with them. Clipping is intentionally used because in many cases it is more transparent than a limiter, but this cannot be generalized to music with soft, pure sounds or human voice.

In waveform view, clipping looks like the top of the waveform was sliced off. It must be evaluated fully zoomed in on multiple loud sections. If the wave is scaled to fit in the range with a limiter, and has normal rounded peaks, then this is not clipping, and cannot be fixed.

In spectral view, clipping looks like a line at the top of the graph for each clipping event. Sustained clipping merges these lines into a band of high-frequency energy, which obstructs features in the original signal.

Unlike compression, clipping can be undone quite sucessfully by extrapolating the signal shape from surrounding data. But the tools for that are not free (free ones aren't very good). The best are Stereo Tool's declipper, and Izotope RX Elements.

For a spectral analyzer I like to use SoX. It is a free command line tool that can generate spectrograms and save them to disk for later viewing. The advantage is that I can scroll through the spectrogram quickly on a slower computer without waiting for it to be rebuilt, while looking a the waveform in another editor. Also, the zoom level can be fixed, which allows to compare spectra of multiple versions of a recording, quickly switching between them. Modern audio editors often have free zoom, and two windows are hard to align. SoX can be added to customizable front-ends, such as Frontah, to avoid using the command line.

Seeing fast amplitude changes on the spectrogram is not easy. To hear an exagerrated effect of a limiter, you could listen to the side/difference channel, subtracting left from right.



Re: Questions on "loudness wars" and/or bad CD engineering - doing a FLAClibrary

Reply #5
...
I've read the term "pump" being used here and elsewhere and it certainly describes the gawd-awful artifact that I find so irritating, but I don't know enough of the vocabulary to identify it in a way others readily understand.

Is there a primer on the common vocabulary used for this stuff?

Thanks for your patience!

Ex recording/mixing engineer here. I wrote an article on this with some definitions and a few listening samples at:
https://www.computeraudiophile.com/ca/ca-academy/dynamic-range-no-quiet-no-loud-r643/

While folks have made references to declipping tools and can help, the sad reality is that over compression/limiting can occur at all stages of recording, mixing, mastering... Even on every individual multi-track channel, going to tape, plus tape compression, and coming back from tape for mixing, each with different compression ratios, attack, release times, plus on sub groups, plus on the overall 2 channel mix, let alone mastering or remastering... Impossible to undo the damage done... The best one can hope for is an older recording that has not succumbed to the loudness war, but if it is new material... Good sounding quality recordings, mixes and masters have become a lost art over the past 25 years :-(


Re: Questions on "loudness wars" and/or bad CD engineering - doing a FLAClibrary

Reply #6
In terms of loudness war the situation is not completely hopeless.
http://productionadvice.co.uk/blog/
The implementation of loudness normalization in various streaming platforms discourages loudness war. It will not change the industry instantly, just wait and see.

About declippers... it's easy to visually smooth out clipped waveform but whether it really sounds better or not is another thing. Unintentional and single instrument clipping is slightly easier to fix but intentional distortions caused by loudness war mixing techniques are basically hopeless. 

Re: Questions on "loudness wars" and/or bad CD engineering - doing a FLAClibrary

Reply #7
Clipping is hard to hear in a busy mix, such as in most rock music and modern pop, which is harsh due to multiple causes, chosen samples, distortion plugins, treble boost. Contemporary music is indeed not good sounding. I guess declipping shouldn't be applied by default. If the samples and total mix takes the bite of the clipping into account, restored tonal components and reduced click, for example, in drums, could radically change the balance.

I looked at Day & Age from the Killers, and there further processing and filtering has been done, and clipped samples cannot be isolated.

I will post clips from real albums that are significantly improved. Stereo Tool expands drums more, and removes all buzzing in the queen sample. It gives better results with heavy clipping. On voice and tones both tools work well, and I can't tell a difference.

In analysis, Izotope RX offers a "multi-resolution" spectrum mode, which gives good frequency resolution for bass and good time resolution for highs on the same picture. Log scale is quite usable in this mode. But it is so extremely slow that I rarely use it. An accidental roll of the mouse wheel causes the zoom to change, and the spectrogram to be rebuilt. With non-integral zoom ratios, the waveform looks fuzzy.

Re: Questions on "loudness wars" and/or bad CD engineering - doing a FLAClibrary

Reply #8
Thanks for the samples. The next part is "unmastering" and that reminded me of this thread:
https://hydrogenaud.io/index.php/topic,113240.0.html

The methodology is somewhat controversial but still quite interesting.

 

Re: Questions on "loudness wars" and/or bad CD engineering - doing a FLAClibrary

Reply #9
Thanks to all for your thoughtful responses.  I will follow up on all links and suggestions. 
My concern is not so much that there is some way to recover the damage done, and I totally agree the artifacts are generally in the original CD media.  I just cannot believe there is any legitimate "mystery" over the reduction in music sales revenue given the overwhelming proportion of literally nauseating product being foisted on consumers.  My aim is really to establish some facts to use in educating oblivious listeners when the opportunity arises.  In a wine analogy, it's like producers have conned most consumers into buying nothing but Ripple.

PS. I listened to the 2014 "Lights Out" album by Ingrid Michaelson.  It is the "pumping" or "ducking" equivalent of The Killers' "wall of noise" technique.  :) 

 
SimplePortal 1.0.0 RC1 © 2008-2018