Remastering (Improving) your own Audio Albums as Big F.Y. to corporate Loudness War trend.

[LF78]

What is interesting about Californication is that multiple different versions do exist.
https://www.rhcpsessions.com/californication-variants

So finally I managed to listen to Californication "rough mixes" and WOW! So much better. Thanks for the tip!

 

[Tool]

I also recommend (2008) Death Magnetic Guitar Hero version. Thank You Guitar Hero.

 

[eliash]

True, but only due to the technical limitations of vinyl. The grooves & needle excursion can't handle a lot of bass, which is why the RIAA curve cuts the bass & boosts the treble when cutting the record. Then on playback it boosts the bass and cuts the treble to restore flat frequency response. This makes it sound better for 2 reasons: (1) keeps bass levels within the groove limitations, and (2) improves S/N ratio in the treble where our hearing is more sensitive to it. Yet even notwithstanding this RIAA curve, records still are limited in the amount of bass. Exacerbating this, typical music has most of the energy in the bass, with energy dropping by roughly 6 dB per octave as frequency increases. Rock/pop is even more so. So when they cut the LP they have to reduce the amount of bass, which reduces average levels, which improves the DR ratio because most of the energy is in the bass.

Overall it can sound better simply because the digital mix had too much bass and dynamic compression to begin with. Yet I suspect it's better by accident, not by intent. They do it because they have to, not because they want to: they do the right thing for the wrong reasons.

Yes, good summary, Vinyl is velocity limited and the max. record levels with no distortion are also frequency dependent (as Shure has shown for their carts in their famous tracking vs. frequency diagrams for the V15 series). Also I buy into the below that you can either have bass or loudness focus on Vinyl, as well as average records´ energy vs. frequency drop.
When removing some bass for Vinyl mastering though, the average level in relation to the previously dominating bass peaks would probably rise, resulting in a lower DR or P/L ratio?! On the Clarity audio meter you can watch the opposite effect, everytime when a loud (bass) passage occurs, the displayed peak level rises and the P/L-algorithm also increases the momentary P/L value accordingly.

Thinking over the quantitative relationship between measured P/L differences and personal dynamic impression comparing Digital and Vinyl, this may not 100% correlate in every case...there seem to be more effects at work (more elaborate compression techniques or only ongoing gain control during mastering?...)

Anyway, I guess that the applied compression (or even "dozen samples-wide" clipping or hard limiting as I have noticed in the "Khmer" CD waveform analysis - which fortunately seems to be a taboo on Vinyl?!) is still the dominating factor for the reduced P/L in digital formats...

 

[MRC01]

...
When removing some bass for Vinyl mastering though, the average level in relation to the previously dominating bass peaks would probably rise, resulting in a lower DR or P/L ratio?! ...

The opposite. Bass makes up most of the total energy, so when you reduce the bass it reduces the total recording level. Typically, they'll adjust so that peaks remain near 0 dB. Now the RMS/average level has dropped while peaks remained the same, so it measures having more dynamic range.

 

[eliash]

The opposite. Bass makes up most of the total energy, so when you reduce the bass it reduces the total recording level. Typically, they'll adjust so that peaks remain near 0 dB. Now the RMS/average level has dropped while peaks remained the same, so it measures having more dynamic range.

So what you are saying is that previously low DR was caused mainly by the strong Bass play itself, but not not fom the underlying rest?!
To be honest, I am still puzzled...Reference levels to calculate P/L and DR are a bit different, I remember DR works on 100% peak and above 80% average levels, while P/L looks at deeper levels for the average...

 

[MRC01]

So what you are saying is that previously low DR was caused mainly by the strong Bass play itself, but not not fom the underlying rest?! ...

The overall level comes from everything: bass (say, below 200 Hz) and "the underlying rest": all the higher frequencies. But it's not equally distributed; most of the energy is in the bass. So if you reduce the level of bass, the overall level goes down almost the same amount. But if you reduce the level of treble, it has very little effect on the overall level.

Of course, this depends on the frequency distribution of the music. You can record test signals or other sounds that don't follow this rule. But it's true most of the time, with most music both electronic and acoustic. If you do a spectrum analysis of most music, you'll see peak energy in the bass, then dropping about 6 dB / octave, more or less depending on the music & recording. A picture's worth 1,000 words: here's a track taken from my collection at random.

 

[eliash]

The overall level comes from everything: bass (say, below 200 Hz) and "the underlying rest": all the higher frequencies. But it's not equally distributed; most of the energy is in the bass. So if you reduce the level of bass, the overall level goes down almost the same amount. But if you reduce the level of treble, it has very little effect on the overall level.

Of course, this depends on the frequency distribution of the music. You can record test signals or other sounds that don't follow this rule. But it's true most of the time, with most music both electronic and acoustic. If you do a spectrum analysis of most music, you'll see peak energy in the bass, then dropping about 6 dB / octave, more or less depending on the music & recording. A picture's worth 1,000 words: here's a track taken from my collection at random.
View attachment 32585

Thanks, very suitable graph to explain and no contradiction from the above as I understand.
When you reduce the bass only, the top peaky blue area above will be aligned to the right located rest (e.g. in peak levels), so the signal will loose a bit of its level diffentiation or distribution width...resulting in a lower P/L or DR from my perspective...

BTW., I found the paper where the P/L and DR background is explained (the DR method is mentioned on page 3):
https://www.tcelectronic.com/_ui/responsive/common/images/lra-design.pdf

 

[MRC01]

Thanks, very suitable graph to explain and no contradiction from the above as I understand.
When you reduce the bass only, the top peaky blue area above will be aligned to the right located rest (e.g. in peak levels), so the signal will loose a bit of its level diffentiation or distribution width...resulting in a lower P/L or DR from my perspective...
...

Usually the signal will GAIN level differentiation: higher dynamic range, the quiet parts will become quieter relative to the peaks.
Remember, the above graph is time invariant, just showing overall energy per frequency.
For example, take a typically heavily compressed modern rock recording, looks like this. It measures DR6:

Now reduce bass levels by applying a high pass filter, 6 dB / octave starting at 300 Hz. Now the waveform looks like this -- note it has significantly better dynamic range. It now measures DR9 -- even though all we did was reduce bass levels (which you'd have to do to cut it to LP).

For comparison, if you apply a low pass filter to the original, 6 dB / octave starting at 2 kHz, the waveform looks essentially the same, and it still measures the same DR6.

In short: when you reduce the amount of bass, it typically increases dynamic range. Especially when starting with dynamically compressed bass-heavy music (most rock/pop). That's usually why LPs measure slightly better dynamic range -- because they had to reduce the amount of bass in order to cut the LP. Which often makes it sound better, since so many modern recordings have artificially boosted bass.

 

[eliash]

Usually the signal will GAIN level differentiation: higher dynamic range, the quiet parts will become quieter relative to the peaks.
Remember, the above graph is time invariant, just showing overall energy per frequency.
For example, take a typically heavily compressed modern rock recording, looks like this. It measures DR6:
View attachment 32625
Now reduce bass levels by applying a high pass filter, 6 dB / octave starting at 300 Hz. Now the waveform looks like this -- note it has significantly better dynamic range. It now measures DR9 -- even though all we did was reduce bass levels (which you'd have to do to cut it to LP).
View attachment 32626
For comparison, if you apply a low pass filter to the original, 6 dB / octave starting at 2 kHz, the waveform looks essentially the same, and it still measures the same DR6.

In short: when you reduce the amount of bass, it typically increases dynamic range. Especially when starting with dynamically compressed bass-heavy music (most rock/pop). That's usually why LPs measure slightly better dynamic range -- because they had to reduce the amount of bass in order to cut the LP. Which often makes it sound better, since so many modern recordings have artificially boosted bass.

Very interesting experiment, which I do not question at all. So the bass spectrum was dominating the DR measurement, due to its own level distribution from my perspective...what I can do is similar tomorrow, with the P/L meter and see if the effect is the same...

 

[Bounce44.1]

The kick and the bass guitar are usually panned to the center to share the load between the two speakers in a typical stereo set up. Everthing else is high passed one at a time until just before the sound is impacted. Sometimes everything below 200hz is combined to mono also.

As far as the lows being compressed it is common to set the compressor on the busses to ignore or not process the low frequencies with the built in high pass filter and depending on the tune it is set to ignore the kick-at about 60hz- or to include the low bass guitar - up to 100hz or so.

The dynamic range is established mainly by the vocals which is also centered as the level is usually vocal, snare, kick with everything else balanced and panned left or right.

Todays music demands the vocal always to be out front which means that the dynamic range has to be controlled by compression and expansion to achieve this. Hence less dynamic range. Unfortunate.

 

[eliash]

Usually the signal will GAIN level differentiation: higher dynamic range, the quiet parts will become quieter relative to the peaks.
Remember, the above graph is time invariant, just showing overall energy per frequency.
For example, take a typically heavily compressed modern rock recording, looks like this. It measures DR6:
View attachment 32625
Now reduce bass levels by applying a high pass filter, 6 dB / octave starting at 300 Hz. Now the waveform looks like this -- note it has significantly better dynamic range. It now measures DR9 -- even though all we did was reduce bass levels (which you'd have to do to cut it to LP).
View attachment 32626
For comparison, if you apply a low pass filter to the original, 6 dB / octave starting at 2 kHz, the waveform looks essentially the same, and it still measures the same DR6.

In short: when you reduce the amount of bass, it typically increases dynamic range. Especially when starting with dynamically compressed bass-heavy music (most rock/pop). That's usually why LPs measure slightly better dynamic range -- because they had to reduce the amount of bass in order to cut the LP. Which often makes it sound better, since so many modern recordings have artificially boosted bass.

Before playing and analysing the P/L in realtime, I have generated a modified CD version of the a.m. NPV´s "Khmer" album, because it contains not only a lot of bass, but also clipping/hard limiting.
As a first thing I compared the DR of both versions and yes (to my explicit recognition, good you brought up this point here) the difference is impressive, have a look (left original CD version, right modified 300Hz-6dB/octave high-pass-filtered)

Especially intersting is the fact that some of the clippling "survived" the filtering process and still the DR is significantly improved, e. g. track 2, from DR9 towards DR14, which is a lot...
HP filtered tracks 3 and 5 show a higher peak value, I am not sure where this comes from, I did no resampling and the Audacity HPF with 6dB slope should not produce ringing...funny

 

[eliash]

Before playing and analysing the P/L in realtime, I have generated a modified CD version of the a.m. NPV´s "Khmer" album, because it contains not only a lot of bass, but also clipping/hard limiting.
As a first thing I compared the DR of both versions and yes (to my explicit recognition, good you brought up this point here) the difference is impressive, have a look (left original CD version, right modified 300Hz-6dB/octave high-pass-filtered)

View attachment 32652

Especially intersting is the fact that some of the clippling "survived" the filtering process and still the DR is significantly improved, e. g. track 2, from DR9 towards DR14, which is a lot...
HP filtered tracks 3 and 5 show a higher peak value, I am not sure where this comes from, I did no resampling and the Audacity HPF with 6dB slope should not produce ringing...funny

In additon to the above, here are the results from the P/L meter measurement.
Same story, here are the filtered tracks´results (measured in a group per vinyl side contents, to compare to my a.m. first vinyl statement about the vinyl P/L measurements)

Peak-to-Loudness (all measured with EBU128 compliant meter on the analog side / ADC: 96K sample rate)

....................Side A..............Side B
LP..................17..................17.2
CD................14.6................14.9
CD (HPF).....18.7................17.7

The HPF tracks per side show even more P/L than the vinyl (even if this doesn´t sound as good as the vinyl, nor as the CD)!
Astonishing, what the bass content is able to produce in terms of DR and P/L.

So the P/L or DR related judgement on music has to be really seen in a more differentiated way...a valuable insight.

Thanks for the fruitful discussion!

 

[Tool]

In additon to the above, here are the results from the P/L meter measurement.
Same story, here are the filtered tracks´results (measured in a group per vinyl side contents, to compare to my a.m. first vinyl statement about the vinyl P/L measurements)

Peak-to-Loudness (all measured with EBU128 compliant meter on the analog side / ADC: 96K sample rate)

....................Side A..............Side B
LP..................17..................17.2
CD................14.6................14.9
CD (HPF).....18.7................17.7

The HPF tracks per side show even more P/L than the vinyl (even if this doesn´t sound as good as the vinyl, nor as the CD)!
Astonishing, what the bass content is able to produce in terms of DR and P/L.

So the P/L or DR related judgement on music has to be really seen in a more differentiated way...a valuable insight.

Thanks for the fruitful discussion!

Can I ask you to send visual snapshot one of the same song from before and after filter pass? Because if Dr improve and compression stays the same, I don't think there is much of sound improvement, and that's the whole point of remastering songs in first place... or do you think there is any better sound?

 

[eliash]

Can I ask you to send visual snapshot one of the same song from before and after filter pass? Because if Dr improve and compression stays the same, I don't think there is much of sound improvement, and that's the whole point of remastering songs in first place... or do you think there is any better sound?

I think I just need to give a link to another forum (I needed more explanation...), the issue there had started 2015 and again actually ongoing with pics, as you asked for:
https://forums.stevehoffman.tv/threads/more-bass-reduces-measured-dynamic-range.472467/page-6
It also shows the effect on filtering a clipped signal, increasing the level. I observed this as well, but did not attenuate before filtering, as I should have done...
I hope this is OK for this purpose...

 

[Bounce44.1]

Can I ask you to send visual snapshot one of the same song from before and after filter pass? Because if Dr improve and compression stays the same, I don't think there is much of sound improvement, and that's the whole point of remastering songs in first place... or do you think there is any better sound?

Ahhhh the crux of the bisquit. First let me don my protective armor specifically protecting the groin area.

If the word "improving" in the title is meant to imply "sound better" perhaps we should claim blasphemy immeadiately as this requires the use of the flawed and inaccurate human hearing as the test apparatus.

Or it could mean that the tools used, analysers-filters-declippers etc, do a much better job than the lowly Mastering Engineer who relies mainly on their hearing and experience to finesse the 2 track to produce the best performance possible which it is capable of.

Both seem to come back to having to use our totally inept hearing to make a judgment.

Or we can agree that there are great and not so great Mixing and Mastering Engineers as there are great and not so great test results.

 

[MRC01]

A little processing (dynamic compression, EQ, etc.) can enliven a recording. And it doesn't have to be this weasel-word "enliven" which is totally subjective. There are technical, empirical and psychoacoustic justifications for this processing, to a certain extent. Even live acoustic music sounds different depending on the venue so there is room for recording engineers and musicians to engineer a certain sound. But how far does it go before it crosses the line into creating more problems than it solves?

This question is nothing new. When RCA Victor came out with "Dynagroove" in the 1960s, it triggered a similar debate. But Dynagroove was mild compared to what is going on now. Give people more powerful tools and some will use them for greater good, others for greater evil. When the DR14 scale measures less than 9, each track has hundreds of instances of clipping, we're well past that line. At that point they're not preserving music, they're destroying it.

 

[captain paranoia]

Give people more powerful tools and some will use them for greater good, others for greater evil.

Ah, the old 'with great power comes great responsibility'...

 

[eliash]

I need to redo the above.
Thinking over and looking at the DR table above shows some tracks where the DR hasn´t changed much. These are tracks (4, 6, 8) which are not clipped at 0dB before and after the HPF.
Unfortunately I cannot do it (with 6dB attenuation before filtering) in the next 2 weeks.
On the other hand I do not doubt the general effect, as can be observed in track 1...and the following vinyl and unaltered CD P/L measurements are fine as well.

 

[Tool]

Usually the signal will GAIN level differentiation: higher dynamic range, the quiet parts will become quieter relative to the peaks.
Remember, the above graph is time invariant, just showing overall energy per frequency.
For example, take a typically heavily compressed modern rock recording, looks like this. It measures DR6:
View attachment 32625
Now reduce bass levels by applying a high pass filter, 6 dB / octave starting at 300 Hz. Now the waveform looks like this -- note it has significantly better dynamic range. It now measures DR9 -- even though all we did was reduce bass levels (which you'd have to do to cut it to LP).
View attachment 32626
For comparison, if you apply a low pass filter to the original, 6 dB / octave starting at 2 kHz, the waveform looks essentially the same, and it still measures the same DR6.

In short: when you reduce the amount of bass, it typically increases dynamic range. Especially when starting with dynamically compressed bass-heavy music (most rock/pop). That's usually why LPs measure slightly better dynamic range -- because they had to reduce the amount of bass in order to cut the LP. Which often makes it sound better, since so many modern recordings have artificially boosted bass.

It's just a trick and miss the whole point... (sorry but i'll quote my self):

I defiantly agree that the whole process hasn't got 100% positive impact, but the whole point at least for me it's not about dr number, it's just an effect of process than goal. It's important to let in some "air" in to the channel and have as much free space between the channels as possible, and finished as much cut pics as possible without damaging the sound. And of course there is always a compromise you have to agree..., but I don't have much choice between bitching about how the sound of a record is not acceptable or try to fix it for my self, as i feel and like it. It works for me doesn't have to for others.

Dr numbers doesn't mean anything in the subject .

 

[Jim B]

WOW, WOW, WOW. This is really interesting to me. To see what computer programs you guys are using to fix your digital music up!!!
I am going to start investigating let me know what your favorite programs are.

This is going to blow your mind but I started listening to music when they played real instruments and nothing was digital and it still had some life to it.

But what I have been doing as a hobby, is I take the poor sounding digital music and re-record it on 10” reel to reel. And then the dynamic range is expanded (DBX) and the bass boosted as well. It’s really been fun and enjoyable!

You know it took more than 10 years after the introduction of cd before they could even come close to the resolution of the reel to reel master tapes.