DIY Dynamic Decompression of Music Tracks

As it turns out, using the multiband expander Pro-MB, the settings found in post #11 is a good starting point for foobar2000 playback in real time. The immediate improvement in highs and restoration of attenuated lows (the areas most often affected by mastering activities--to be undone), even if in very small increments, can dramatically increase the sense of presence/realism of the music tracks--without having to previously run the tracks through an audio editor and saving them. This is exactly like using an add-on equalizer, like the one that's found in foobar2000, except now you can control dynamics and its flip side: damping. (I find that the deleterious effects of using mastering compression includes underdamped percussive instruments/voices that leads to an unnatural sound quality...putting more apparent distance between the listener and the music, and turning the music into pablum).

Real-time tweaking of the settings is fast, easy, and intuitive. I find it useful for opening up the dynamics of older popular music genres dating back to the 1970s (rock, pop, folk, etc.). Some of the most successful applications include those music tracks whose then-prevalent mastering ethic was "subdued highs". The Pro-MB plugin (multi-band expander) can quickly be called up while the music is playing, then a preset imported, while the music is still playing. The music is immediately expanded and EQed--in real time--and the resulting change in amplitude vs. frequency vs. the incoming base track is visible in the plug-in display. This is powerful and easy.

If using the Pro-Q4 plugin (dynamic equalizer), correcting any pre-existing mastering EQ used can be visually adjusted even more easily by watching the dynamics vs. frequency as shown in a peak spectrogram underlying the dynamic EQ settings. So setting and adjustment of both EQ and dynamics vs. frequency are done easily--simultaneously. Refer to the YouTube video in post #19, above to see the plugin in use in real time using foobar2000 as a player.

Any player that can run VST plugins can use these two dynamics expanding plugins directly, without having to first edit and then save them.

On the flip side, if you really like the resulting sound from current plugin settings, you can easily save the expanded version of the track right there so that it can be called up directly in the future without having to turn on the plugin and recall the saved settings.

Chris
 
Last edited:
  • Like
Reactions: MA043HA
This is exactly like using an add-on equalizer, like the one that's found in foobar2000, except now you can control dynamics
I've been wanting such a facility for a long time. I even once imagined contacting the "Cockos" people about refining their multiband compressor VST plug-in to better accommodate expansion, perhaps a version strictly for audio playback. For some reason, their band controls were allowed fractional ratios, leading to expansion possibility, though not much control "surface" was given between zero and one. Almost like whomever programmed it dropped in a zero as the minimum ratio instead of one.

Glad to read someone else has done it; thank-you for sharing your experience here where I can see. Looks like I downloaded ffpromb131x64, which I can add to the dozens of other things I'm looking forward to fool with during cabin fever time.. I've got a new toy coming for Xmas; a Soundcraft UI24, which will probably soak up a little time too. I bought it for myself, as there's no one else alive who could could imagine I'd even want such a thing.
 
  • Like
Reactions: MA043HA
I've been using Fabfilter Pro-MB for some time now to reverse some of the effects of music track dynamic compression during mastering...but not mixing, of course. I've found some interesting tidbits of information while using it:

1) Dynamic expansion with a good multiband upwards expander (like Pro-MB) is possible if the dynamics of the music track haven't been completely crushed into oblivion. In DR Database terms, this means that if the music track has at least a 6-8 dB rating using the "TT Dynamic Range Meter" (the standalone version or as a plugin for foobar2000), you have a reasonable chance of a good listenable outcome using multiband expansion. The maximum DR ratings that can benefit from expansion seem to be around 13-14 (dB-crest factor) on the DR Database scale. Any tracks compressed to lower dynamic range values seem to be "lost causes". It's much more productive to go on a hunt for a higher dynamic range version of the recording rather than trying to use a multiband expander.

2) The plugin settings used to expand tracks are usually most successful in increasing dynamic range in the higher frequencies of a track than the lower frequencies below ~100-200 Hz. This means that the crest factor ratings of the tracks from the TT DR Meter may not change very much (due to the fact that crest factor is really measuring bass dynamic range, but not really higher frequency DR), but the tracks themselves will present a much more crisp and forward sound quality for dynamic transients--like ride cymbals, crash cymbals, bells/cowbells, glockenspiel, all handheld percussion instruments (particularly latin instruments), string attack transients, marimba/vibe strikes, drums, and human voice transients.

3) The "steeliness" of string orchestras used in lush pop string arrangements and mass string orchestral scores (classical) will largely be suppressed if using the multiband expander carefully. This was a big surprise.

4) Any human voices, particularly female voices, will begin to sound much more realistic and without typical harshness that comes with listening to recordings having compression applied during mastering. This was the biggest surprise of all in my explorations using the expander.

5) The albums that respond most strongly and easily to multiband expansion seem to include those that had analog compressors applied during mastering (e.g., early-late 1970s albums) CBT test training, which apparently used continuously varying nonlinear compression curves without a sharp knee breakpoint/thresholding applied during mastering.

6) The music genres that seem to respond most strongly are funk, rock, folk, jazz (including smooth jazz), progressive rock with lots of drums/percussion, dance, and related genres. Those genres that seem to respond with more difficulty include smoother music genres having lots of midrange energy (i.e., ambient, new age, classical string orchestras, classical guitar, etc.).

7) All tracks that have used multiband expansion need to be adjusted before and after expansion using parametric EQ demastering.

__________________________________________________________________________________________________________________________

It's clear to me that the use of a multiband dynamic range expander is now standard equipment for my typical demastering tasks that I perform on most of my incoming discs.

I don't recommend trying to do expansion "on the fly" (without review and adjustment of expander settings track by track) unless extremely mild expander settings are used, which leads to barely audible subjective differences in the compressed tracks. Using an expander offline and taking a little time to get the setting right to get the best and most audible results is my strong suggestion in using these type of plugins. However, the task only has to be done once for the music tracks--then saved for future listening, instead of having to set up a custom string of plugins and parametric EQ settings each time a music track is played.

(My original text above was originally posted to another audio forum, albeit an unstable one, and has been moved to this forum to ensure its longevity.)


Chris
I know you guys sure have lots of tips and recommendations to do this in a good way. Hardware and/or plug-in-wise doesn't matter.
 
Fixing the compressed dynamics of damaged music is the actual subject of this thread. To help keep this fact centered in the mind's eye, it should be noted that there are no microphone recordings (i.e.,those taken directly from the microphone without intervening processing or modification) that need to be "restored" to make them sound more like the real instruments and voices in-room.

The process known as "enshittification" (wiki: enshittification) is the source of the problems that lead to efforts like the present to restore some sort of original sound quality to tracks that have largely been trashed.

Those responsible include a portion of the music production industry, i.e., largely the mastering houses. This practice is no doubt spurred on by record company A&R executives trying to maximize perceived revenue--at any cost.

What percentage of today's released music is affected by these processes? Well over 99.9%.

And well before music streaming became a thing, the same process of enshittification of music was alive and abundantly practiced, as evidenced by older recordings found on hard media (CDs in particular, starting in earnest in 1991 with the introduction of multiband digital compressors). A few practicing mastering professionals began to decry these increasingly intrusive practices, as evidenced by the "Loudness War" articles in the mid 2000's.

These practices have variously advanced themselves (usually identified within certain music genres, in particular) into the popular jargon and even expectations of customers, just like the progression of drug addiction in those countries where no laws are in place to check the widespread addiction.

So for those audio enthusiasts that believe the quality of the recorded music they play is based on some standard that preserves the essential sound quality that was captured by the microphones-- think again.

You might think this is an old story. It is. The problem is that nothing has changed--no movement toward standardization of sound quality in produced music recordings has been made over the 20+ years of crusading against these practices.

So why a thread on ways of increasing dynamic range of damaged recordings?

It should be obvious by now.

Chris
 
Last edited:
Fixing the compressed dynamics of damaged music is the actual subject of this thread. To help keep this fact centered in the mind's eye, it should be noted that there are no microphone recordings (i.e.,those taken directly from the microphone without intervening processing or modification) that need to be "restored" to make them sound more like the real instruments and voices in-room.
As with all "drugs", therein lies the potential for abuse. I'd argue that any recording system is going to compress some, just in getting the pressure fluctuations from the air through a physical barrier and electronics, onto some medium that can be played back.

Albeit we've all been trained on compressed music, with some of the finest examples of tube based compression hardware originally designed to help keep up modulation levels of radio station transmitters, there is a valid use case for it in live performance.

Example, I perform regularly at a coffee shop open mic. The two fellows running it know nothing of compression, let alone ambient effects. They turn up my voice so loud I cant hear my acoustic guitar at times and I find the dynamics of it distracting from what I'm trying to do. In a similar way Neil Young didnt like the clicks of a photographers camera happening while he was playing.

As if it's not hard enough to sing in key floating that over the top of playing chord changes in time, now I have to consciously "ride" the level of my voice to emulate vocal compression myself. Else, I can easily blow people right out of their chairs should I slip up on that via the huge volume transients that can result.

I'm sure an audience wants a pleasant experience when attending something like this. While dynamic range is exciting, it can also at times frankly get on your nerves. As far as ultimately sounding more toward a highly "processed" or "mastered" final mix when playing live, I for one dont seem to mind at all. You see in live music, you're "straight" from mic and guitar pickup to the PA speakers; no recording medium at all offering any consequent limit on SPL dynamics.

Just what one would think you'd want, well, maybe for the symphony orchestra' sound reinforcement.

I've recently purchased a Soundcraft ui24r mixer. I bought it mostly for the any compression curve you want on every channel. Sometimes I practice mic'd and guitar, listening via headphones. I dial up the steepest compression possible, to make my practicing more pleasant. Not the first time either, practicing in a rock band context maybe 40 years ago was so deafening, so fatiguing due to SPL dynamics (think acoustic drums and guitar amps in a room) I couldnt continue.

So I put the drums behind a wall, and put in a window, micing them behind there. Everything else direct to the 12 channel board. DBX compression at the mixer output. Everyone monitoring using headphones after compression. Whew! That's what literally made it possible to continue regularly. To those that do it the garage band way, I honestly dont know how you can physically keep it up.

So compression to me is one of those "is and it isnt", "both and", cant live with and cant live without type of things. It has its place, can certainly be abused, but taking it completely away just might not give an all that great listening experience as well. Of course, there's music genres where "pleasant" is definitively NOT what they're going for; they like a few "pain" spikes along the way and SPL transients are certainly capable of providing that.

Like everything in Audio, ultimately depends on your cuppa tea!
 
Last edited:
If expansion was a magic bullet then it would be used everywhere. It's not.

And if you listen to things produced past 1990's this is a lost cause as you will never have the possibility to undone what have been done through multiband dynamic processing. In fact you can't recover a dynamic treatment, once it's applied original dynamic is lost ( transients have been modified).
So true!
 
I'd argue that any recording system is going to compress some, just in getting the pressure fluctuations from the air through a physical barrier and electronics, onto some medium that can be played back.

Live performers don't need compression, nor any of my loudspeakers or electronics, nor any of the recording formats I use.

So what needs recording compression? It certainly isn't the human hearing system.

I've found empirically that recordings having a crest factor of 17 (dB) or greater sound pretty much like the real thing (and of course, not all recordings have the instrumentation to produce that high of a crest factor). If the crest factor is significantly below this level, I've found that it will not sound anything like a live performance.

How many of your reference recordings have a crest factor of 17 or greater?

Chris
 
Last edited:
How many of your reference recordings have a crest factor of 17 or greater?

Just a couple. I've got some direct to disks from the 80's that probably do, although never measured them. They do sound great and have an air to them. Pity given today's absolutely incredible dynamic range possibilities that they squeeze it down to 3 bits.
 
Just a couple. I've got some direct to disks from the 80's that probably do, although never measured them.
Do state the measurement process. Can you import the music file into Audacity and get this value? Other software can do the total frequency spectrum across an entire recording, is there one that likewise can do this crest factor measurement? Last time I heard that term was in AC mains distortion.

So what needs recording compression?
Maybe people's brain? As I said, I've been listening to compression and its artifacts since the day I was born, almost 70 years ago. "You dont listen to much live music then". For sure no, not in an hour-by-hour comparison to the compressed stuff, which I assume was every radio, every TV, every vinyl record, every cassette...

It could be for most people, that's what sounds "right" or "normal". If you want to sell something to them... Also I postulated for certain situations, while perhaps not needed, its a useful audio tool for creating a relatively stress free listening environment.

It would be an interesting experiment in that context, across venue owners providing different genres of live music, how their clientele would respond to setting their PA dynamics to provide a crest factor ">/=17 or <<17" and do a study. Probably could wring a masters or PhD out of that one.

If you want your sound system to sound different than most anyone else's, for sure do whatever you realistically can to "glow-up" the existing recording more toward the dynamics you feel it should have. I'd practice that, wherever it worked, if only to not have the same old, same old thing. Glow-up being a term used for when a drug addict becomes sober - and the light in their eyes coming back on.
 
Maybe people's brain? As I said, I've been listening to compression and its artifacts since the day I was born, almost 70 years ago. "You dont listen to much live music then". For sure no, not in an hour-by-hour comparison to the compressed stuff, which I assume was every radio, every TV, every vinyl record, every cassette...
This is the definition of "circle of confusion", unfortunately.

It could be for most people, that's what sounds "right" or "normal". If you want to sell something to them... Also I postulated for certain situations, while perhaps not needed, its a useful audio tool for creating a relatively stress free listening environment.
If I had a cookie for every time someone tried to justify the status quo by such means, I wouldn't need to buy any more cookies ("bisquits" in the UK).

It would be an interesting experiment in that context, across venue owners providing different genres of live music, how their clientele would respond to setting their PA dynamics to provide a crest factor ">/=17 or <<17" and do a study. Probably could wring a masters or PhD out of that one
I never said anything about amplified live music. PA systems typically are assembled based on lowest cost gear. I think a lot of people associate the term"PA" with low quality (which it doesn't have to have low quality, actually).

"glow-up"
Where did you find this term? I wouldn't use it.

If you like your compressed music because you imprinted on it at an early age (such as jukeboxes, etc.) and apparently didn't listen to anything else, that's up to you. That 's not my background, however.

Chris
 
I did a few songs off the one, Amanda McBroom Growing up in Hollywood. ACX reported -10/-30, -2/-26, -6/-31, -6/-28, -2/-28 peak/rms on the first few songs. Those are reported in db. I think crest factor in db for the first pair would be 20 (-10 - -30)in db, or a crest factor of 10. The largest pair 2/28 would give 26db or a crest factor of 19 I think.
 
I never said anything about amplified live music.
Actually I wish I could just play on stage without it, maybe I'd learn to sing "right". But, the venue has a pair of EV powered speakers on stands and a Behringer mixer, so the folks running the show feel the sound needs to be reinforced for some reason. For the 10 - 15 people in the place...

The venue felt they needed such equipment on hand as an attraction to come make use of - and forked out the cost of mixer, mics and speakers. Behringer felt adding a compressor knob on 4 channels would sell mixers. All part of this circle of confusion, among common people and their audio / listening experience imprints.

I try to squeeze myself into the available format and do the best I can with it. The guys doing sound clearly have their egos; I'm not about to step on their effort by telling them you should be doing it this way or that. Without them, there wouldnt be an open mic and the fun I have attending.

Another format would be two mics hanging from the ceiling over the stage choir style and the PA goes as loud as it does before feedback - and that's all the additional SPL you get, albeit with higher dynamic range. Versus lines in from electrified guitars and pedal boards and close mic'd vocals, which can get a lot louder.

I donated a Roland electric piano I stumbled upon at a yard sale for practically nothin'. It sounds compressed too, compared to a real piano, but - nice enough. Of course I'd like the sound of a real piano there, but that would be hundreds or thousands to provide...

My only assertion is compression of audio dynamics has its place, depending on the situation. I agree that it's too bad that practice has leaked into everything, and to the point where it's inextricable.
 
The real problem that I see is "one size fits all" mentality in released music by record companies, noting that the mixdown tracks usually retain much higher crest factors (and overall fidelity) than the released versions, but require no extra effort.

In fact, they are actually "free" since these are the same tracks provided to the mastering houses before they do their work. Those mixdown tracks can easily be offered via download sites at essentially no increased cost. But I still find that "one size fits all" seems to dominate industry thinking, and a concerted effort by record companies to charge MORE for DSD backup copies on SACD rather than less.

I know of at least one record company that does no mastering (only mixdown tracks): AIX. (There may be others.) The fidelity of these recordings is usually very high--especially after their shift to Blu-Ray audio format, replacing DVD-A format.


Chris
 
Those mixdown tracks can easily be offered via download sites at essentially no increased cost.
One time I worked for Amzn, briefly - about a year. They had an internal contest for the "next big thing" open to all employees. The job was one of those sit around in case something breaks, occasionally do some routine maintenance positions, try to improve things left behind by the last guy, undocumented. So I had substantial time to think.

I knew Amzn had the tremendous music system to compete with the likes of Apple. I knew they had gobs of network bandwidth available, streaming countless movies. So I proposed that they deliver 10 tracks instead of two, which the corresponding "app" would handle and mix / produce into a listenable form at the endpoint. The song/album would come standard with the original engineers mix/production, or the listener could mix/produce the tracks any way they wanted as an alternative - however they saw fit with the controls provided. Level, eq, compression et al.

Several in reply to the idea liked it, but not nearly enough interest for the idea to be considered. One comment was you're offending the original engineer / producer and their precious work! I assume by offering the possibility of doing something a little different. While I on the other hand was imagining on-line forums chock full of "my mix (attached) sounds better than yours, having actually tried each!" for some Fleetwood Mac song...

I would assume such a "10 track" delivery would be uncompressed and the compression function would be part of the playback app, which could be altered from the original settings by the recipient if desired.
 
I did a few songs off the one, Amanda McBroom Growing up in Hollywood. ACX reported -10/-30, -2/-26, -6/-31, -6/-28, -2/-28 peak/rms on the first few songs. Those are reported in db. I think crest factor in db for the first pair would be 20 (-10 - -30)in db, or a crest factor of 10. The largest pair 2/28 would give 26db or a crest factor of 19 I think.
Here are the unaltered values for the CD version of that album, as reported by the Dynamic Range plugin:

Growing Up in Hollywood Town - Amanda McBroom Lincoln Mayorga.jpg


foobar2000 1.5.1 / Dynamic Range Meter 1.1.1
log date: 2025-03-02 03:31:40

--------------------------------------------------------------------------------
Analyzed: Amanda McBroom and Lincoln Mayorga / Growing Up In Hollywood Town - from disc
--------------------------------------------------------------------------------

DR Peak RMS Duration Track
--------------------------------------------------------------------------------
DR12 -7.20 dB -24.96 dB 4:09 01-The Portrait - from disc
DR18 0.00 dB -20.17 dB 1:53 02-Peter The Hermit - from disc
DR13 -5.03 dB -26.26 dB 3:11 03-The Rose - from disc
DR18 -0.76 dB -24.76 dB 2:55 04-You've Lost That Lovin' Feelin' - from disc
DR15 -3.35 dB -23.33 dB 3:30 05-Dusk - from disc
DR18 0.00 dB -20.90 dB 4:00 06-Hooray For Hollywood - from disc
DR19 -1.37 dB -24.48 dB 3:38 07-Love Letters - from disc
DR17 -0.09 dB -22.78 dB 4:20 08-Amanda - from disc
DR12 -10.55 dB -27.95 dB 3:01 09-Silent Lady - from disc
--------------------------------------------------------------------------------

Number of tracks: 9
Official DR value: DR16

Samplerate: 44100 Hz
Channels: 2
Bits per sample: 16
Bitrate: 626 kbps
Codec: FLAC
===================================================================

The crest factor for each track is reported by the first column of numbers (in dB) taking the (I believe) second highest peak per track for reference. Note the reported instances of clipping in tracks #2 and #6 (i.e., 0 dB track peak)

Chris
 
Not too far off from my audacity numbers. And mine could have some extra from clicks/pops on the vinyl. I recall buying the album in the 80's and thinking it was very realistic in the dynamics. I don't think sheffield did much processing, probably a final compressor to avoid overmodulating the cutter. It was direct after all. As I said earlier though, it is sad with the DR available with 24 bits that things are squished to near zero DR. The sheffield like all the albums I purchased back then got played very infrequently, the first play being used to record onto a Nak ZX-7 with good tape. And again, sadly that 80's era tech would be more than sufficient to capture today's hi-rez near zero DR recordings. Curious, what is the worst DR you've seen. Does it get to less than 3db?