Can Bad Recordings be Fixed?

[Audiophile Neuroscience]

Can Bad Recordings/ masterings be fixed and made to sound good?

 

Maybe a little creative EQ for each track?  Decompression/Declipping? Maybe a better system does it for you (realizing the potential within) ? Color me warm with sonic sunglasses?

 

This topic was inspired by posts in another thread some of which are quoted below to kick things off.

-----------------------------------------------------------------------------

23 hours ago, PeterSt said:

The problem is that it is quite hard to tell whether a recording is a bad recording. I have too many examples of that seemingly being so, until years later I improved something and suddenly the bad recording became a good recording. Won't count for everything and all of course, but say that in the general scene we both will know (but exclude me from classical) chances are more than 90% that the bad recording is your (our) poor system.

 

23 hours ago, Audiophile Neuroscience said:

 

I can accept that a bad recording can sound less bad depending on what you have tweaked. The risk is you color the sound to suit the recording type or otherwise gravitate to recordings that suit your colored system.If the goal is transparency, true transparency then bad recordings should be heard for what they are, not with sonic sunglasses.

 

OTOH I am exploring various ways of tinkering with bad recordings with some creative EQ or other track specific methods. NOT in making changes to my whole system.

 

19 hours ago, PeterSt said:

 

No, I am not saying that; A bad recording is a bad recording.

But you tell me how you can definitely for 100% sure know that a bad recording is a bad recording ? (with the notice that a "recording" as such is seen by us through its pressing/mix/(re)master).

 

1. What we both will think is that a flat sounding recording is a bad recording.

2. Stuffed ears too few highs is a bad recording.

3. Booming bass is a bad recording.

4. Gray sounding highs must be a bad recording.

 

Ad 1.

Rolling Stones - Get Yer Ya-Ya's out. Turns out not to be flat at all. Somehow things can be done so that it expresses as much depth as the other albums.

 

Ad 2.

Deep Purple - Machine Head. In a very strange fashion the highs are buried. But miraculously a year ago they appeared.

 

Ad 3.

Heavy D. and the Boys - All albums. Were not playable for me. Remove the (system !) distortion from the bass (consider that 100% of people suffer from this) and all appears fine.

 

Ad 4.

Beatles - e.g. Get Back. If you'd hear it over here you'd put a lot of money on the perception of listening to a cover.

 

Don't confuse this with poor remasters which never ever will be OK. But also look at this example (of my experience) :

For many many years I have thought and said that Beatle's Love was about the only well-done remaster I ever experienced (the middle parts of the songs are "orginal"). With all my guts and experience I seriously was convinced that those middle parts of the songs remained untouched, thus equal to the originals, btw including above mentioned Get Back.

This stopped when I created the Clairixa USB cable. From then on, I suddenly heard a huge difference and the Love remasters suddenly were a total failure.

I assume you know it. And I also assume that those who know it, will not agree with me.

 

But what really happened was that the originals got better again, this time by means of an USB cable.

 

A few pages back I gave that list of "old hits", like from the 60's. I am confident that you won't believe me when I tell that almost all in that list sound like they were created today (but without today's compression). But might you believe me, then you won't believe me that 20 years ago this was just a pile of crap while 34 years ago Dire Straits sounded "perfect" through the same playback system (I only compare CD's).

 

The experience with Bill Evans is in a kind of other league because if you don't coincidentally have the version I have (and which appears super rare), it is a "poor recording". This, while if you see me screaming in that link about the (1959 !) quality, you wouldn't understand if you have one of those many other versions, including the totally failed Hires. So :

a. chances are fair that you have that other version and think it is a bad recording which is true;

b. chances are more fair that you think it is all crap anyway because too old;

c. the truth is that such an album sounds so good since a stupid playback software improvement, back at the time.

 

Btw, this thread is about the Lush cable, and with that too you hear people say more than once that old recordings got so much better.

 

17 hours ago, CuteStudio said:

 

 

Perhaps you are mixing (sic) recording with mixing and mastering? Today's problem is at the mastering stage and ruins almost all modern music - and therefore HiFi. Very little escapes unscathed. At the end of the day we can't all listen to good pressings of The Beatles, The Stones, Bowie, Dire Straits etc because some modern music is in fact very good, we just never know due to the damage done when it's mastered into a block on continuous noise.

 

Adele's 19 was not too badly mastered for the time, i.e. it was average to its peers. 21 was worse. 25 is even worse. For many artists like Adele the mastering gets worse over time, Adele just happened to have named hers conveniently to track the deterioration in quality.

The word 'Re-master' really means 'Modern master' which means heavily compressed and about half the time: actually clipped.

 

The way to definitely tell if you have bought a bad master is actually very simple: There are two easy methods:

 

1) Is it a modern pop record? A 99% chance it's been wrecked.

2) Open it in the free Audacity music player. If it looks like a brick shaped object it has been significantly damaged during mastering.

 

There is no possibility of making these recordings sound good by changing your system. The square edges of the clips can only be rounded off by reducing treble, at which point it's not HiFi anymore. There is a small point in your favour which many miss - quite a lot of HiFi makes a bad master sound even worse: so I agree the playback system has some role.

 

With heavy digital processing it is possible to make them quite listenable, but they will NEVER be the quality of a properly mastered track. The worst damage is actually that which creates the 'brick' shaped waveform but doesn't actually clip, because the data of the original peak heights is lost when they are all forced to the same level and that lost data can be vaguely inferred but is essentially the most significant bit of the digital waveform missing. I.e. People argue about 16bit because they worry about the last bit (bit15) not having enough detail but todays digital music has at least 6dB (bit0 !!) missing from it. It's a serious problem and no one will fix it.

 

This article describes the problem: http://www.stylusmagazine.com/articles/weekly_article/imperfect-sound-forever.htm

Bear in mind that article is 11 years old now so quality today is significantly worse that the author is describing, each year quality falls further.

 

Sadly many listen to dreadful sources on their HiFi: Checking in Audacity would shock them from their delusion of 'bit perfect' CD quality sound.... for these reason I suggest that Theresa is correct: the source quality is key.

 

8 hours ago, fas42 said:

Peter is one of the very rare individuals who has also discovered the magic that lies within recordings that nearly everyone else has discarded as not being worth listening to - he is still hestitant about whether all recordings can be "rescued", but he can rest assured that it will be difficult to find anything that can't rise, phoenix like, from the ashes of crap sound .

 

Above he mentioned Deep Purple - Machine Head: I have an original, unfiddled with release - and this is an amazing album - blows all the audiophile "rubbish" I've heard 100's of yards into the weeds . So much to hear in it, with fabulous changes of pace and atmosphere. So, you see, it's very easy for me to appraise some unknown rig - if it can't extract at least some of the magic in this recording then it's stuck in first gear; it's trivial to be able to hear what a system is doing wrong, once you know what to listen for ...

 

6 hours ago, Audiophile Neuroscience said:

 

Hi Peter,

Thanks for the considered reply.

 

I agree with Cute Studio's response.

 

I do take your point about it being difficult to definitely tell what is a bad recording vs just imperfect playback. It is a related issue to how do you tell if your system is neutral. In either case, short of being in the recording studio and comparing the versions, there is no other gold standard experience-based  way.

 

That said, I don't consider it is all that difficult to know that you are moving in the right direction, even if you don't know for sure if you have arrived.

 

If transparency/neutrality is the goal then virtually all recordings/masterings should sound somewhat different. Certainly not all sound the same in any one area be that warm and lush even if that is your taste.

 

You can also get clues by playing recordings on various different playback systems and see if there is a commonality for that recording eg sounding compressed and loud on all systems.

 

As mentioned by others, tracks can be analyzed and deemed poor recordings/masterings based on compression, clipping and other factors.

 

When previously bad sounding bad recordings sound better after a system change I would expect previously good sounding good recordings to also sound better but the impact may be more or less.

 

As for poor playback, again there will be a pervasive quality for ALL recordings.If you have gear reproducing the bass as flabby and loose, lacking articulation, definition, and speed.....it will be that way, more or less, on ALL recordings.If the bass is great on most recordings and poor on some, then relatively the latter are poorly recorded.That is NOT to say that they cannot be further improved. We are talking relativity and direction here, not absolutes.

 

 

It is not exactly unique for most audiophiles to experience/claim that a system upgrade/change has made a universal improvement in SQ.For me at least that equates with a *move* towards transparency and neutrality.

 

I am lucky to have a fine system and have striven for transparency, the system and room getting out of the way of the music. I have been startled at how good some old stuff can sound and for example, cliche or not, have been rediscovering old music from Julie London to Dire Straits. There is a lot of information hitherto not realized on some old recordings. Then, if your system removes every ounce of *superimposed* grain, noise,glare and other artefacts there are additional boosts.

 

I am talking about reaping the benefits of increased transparency and not further detracting with superimposed flaws. Not sonic sunglasses coloring the playback.

 

My prediction is that the best system on the planet *should* resolve and render a bad recording as a bad recording.Best sound will come from best recordings. People who say everything sounds great on their great system I believe are listening to euphonic colored sound, a facsimile of the real thing, not Hi- Fidelity, faithful to source. I have no issue with that, just not my cup of tea.

 

 

[PeterSt]

2 minutes ago, Audiophile Neuroscience said:

Decompression/Declipping?

 

Only that IMO. But theoretically. I have no experience with it.

And, because it will again massively process the file (the music data) I can't believe it will work for real. So the result may be better on the dynamics and such, but the consistency will be gone even more.

 

So I vote No. 

[fas42]

Decompression/Declipping ... yes! The experiments I did showed the potential, and I was merely feeling my way, purely explorative efforts. One pop track was just savagely compressed in the final mixing, and the parameters used in that operation could be deduced quite straightforwardly, from examining the waveform - applying the inverse of those parameters, in editing software that allowed uncompressing, did a nice job - there was no loss, from my POV, in the integrity of the track, but the massaged result sounded very much like a straight mix of the instruments.

 

Beware! There are plenty of utilities and such out there that claim to do things, but my experience of them is not good - I went back to basics, in analysing what could be seen in the waveform, and from that deduced what could be a good approach.

[Fokus]

21 minutes ago, Audiophile Neuroscience said:

Maybe a little creative EQ for each track?

 

Absolutely. The latest Kate Bush live recording sounds horribly dull when played on neutral speakers or headphones. Of all the recordings I have I deemed this one totally unacceptable. So  I have made my own remaster.

 

You can also play with the stereo image and the ambience.

 

21 minutes ago, Audiophile Neuroscience said:

Decompression/Declipping?

 

Less so. Compression cannot be undone without intimate knowledge of the compressor's settings. Any attempt to expand will likely improve matters here, only to mess them further up there.

What could be done is simply backing off levels, so that any subsequent processing has some headroom. But don't expect miracles.

 

There are people who apply high-pass or all-pass filtering to compressed/shredded tracks. Sure, this visibly livens up the waveform, and it tricks calculation tools into extracting a higher DR figure (much as these report higher DR from vinyl, even for the same master). But it does not really change the sound.

[Audiophile Neuroscience]

3 minutes ago, Fokus said:

Absolutely. The latest Kate Bush live recording sounds horribly dull when played on neutral speakers or headphones. Of all the recordings I have I deemed this one totally unacceptable. So  I have made my own remaster.

 

Would you mind sharing a little of how you did this and with what tools?

[Audiophile Neuroscience]

4 minutes ago, fas42 said:

Decompression/Declipping ... yes! The experiments I did showed the potential, and I was merely feeling my way, purely explorative efforts. One pop track was just savagely compressed in the final mixing, and the parameters used in that operation could be deduced quite straightforwardly, from examining the waveform - applying the inverse of those parameters, in editing software that allowed uncompressing, did a nice job - there was no loss, from my POV, in the integrity of the track, but the massaged result sounded very much like a straight mix of the instruments.

 

What DAW did you use?

[Fokus]

5 hours ago, Audiophile Neuroscience said:

 

Would you mind sharing a little of how you did this and with what tools?

 

This one case was a bit funny. I was listening at work, with iTunes, DragonFly, headphones. Totally fed up with the sound I opened itunes' graphic eq and tweaked it until it sounded like music again.

 

Later, home, I approximated my settings using Adobe Audition's parametric eq,  which is my weapon of choice for such problems. Strangely enough, it sounded over-smooth. Eventually I simply copied the iTunes graphic eq parameters to Audition's graphic. The resulting dirtier signal probably just sounds more alive.

 

[Apesbrain]

5 hours ago, Audiophile Neuroscience said:

 

What DAW did you use?

 

Any DAW that supports VST plug-ins should be able do the job.  Audacity and Wavosaur do and both are free.  Each comes with its own set of effects which may be enough to get what you want without bothering with plug-ins.

 

Whichever you choose, go easy (and keep a backup of your original file).  A little bit of enhancement can sound good, but it's easy to pour on too much and get crap.

[Fokus]

Indeed.

 

And if you pour on much of anything, the overall signal gets louder, which sounds 'better' on its own. You have to take some distance, account for the increased loudness, and only then decide if it is really better than the original.

 

[fas42]

15 hours ago, Audiophile Neuroscience said:

 

What DAW did you use?

 

Reaper. Easily the most powerful I've come across, that's available to play with for zero cost.

 

Trying to mitigate the impact of compression without performing an undoing action with a high level of accuracy will be largely useless - all sorts of disturbing anomalies will be introduced. This is an exercise of performing serious analysis of the waveform, finding the patterns which mark the compression operation, determining what parameters were used in the software to create those patterns, and then applying an inverse operation. Using trival methods to try and 'fix' will yield trivial output, of little value for listening to, IME.

 

[fas42]

EQ is useless ... to me. All it may do is reduce the apparent level of annoying artifacts introduced by less than optimal playback. We have all had lousy radios playing with screechy, distorted high frequencies - easy fix, wind down the Treble - nothing's been resolved, we're just making the ugliness less noticeable.

 

Only real solution is to eliminate the ugliness - which is not the recording, but how the playback mishandles "troublesome" content. The attitude to have is that the recording is always fine, but the playback can't cope - "you can't handle the truth!!"

 

The big plus is that the 'difficult' recordings often end up becoming the most rewarding, because the intensity and richness of their content delivers a more powerful, emotional experience.

[Audiophile Neuroscience]

9 minutes ago, fas42 said:

 

Reaper. Easily the most powerful I've come across, that's available to play with for zero cost.

 

Trying to mitigate the impact of compression without performing an undoing action with a high level of accuracy will be largely useless - all sorts of disturbing anomalies will be introduced. This is an exercise of performing serious analysis of the waveform, finding the patterns which mark the compression operation, determining what parameters were used in the software to create those patterns, and then applying an inverse operation. Using trival methods to try and 'fix' will yield trivial output, of little value for listening to, IME.

 

 

I am a great fan of audio and recording/mastering engineer Barry Diament and he also recommended Reaper for me to try some subtle Eq.

 

Generally Barry recommends *as I understand it* often only 1/4 to 1/2 dB changes with *low* Q ( wider shallower peaks) of around 0.66 equivalent to 2 octaves and generally fiddling with less turnover frequencies, which turnover frequencies depending on the result required.

[Ralf11]

Phase Linear Auto-Correlator

 

Plangent Process

 

 

[fas42]

The Plangent Process looks good, in that it undoes the wow and flutter embedded in the analogue master tapes - filling in of "missing frequencies" is a question mark though ...

[Ralf11]

need predictive bit blit for that

 

 

[gmgraves]

20 hours ago, PeterSt said:

 

Only that IMO. But theoretically. I have no experience with it.

And, because it will again massively process the file (the music data) I can't believe it will work for real. So the result may be better on the dynamics and such, but the consistency will be gone even more.

 

So I vote No. 

 

I say that bad recordings can, sometimes, be somewhat improved, if not completely fixed. For instance, many modern pop recordings are heavily compressed. That can be reversed using a com-pander like a dBX, IF (and boy is that a big IF) you know (or could figure out) the compression and release slope that the record company used to compress the album in the first place! Otherwise you're just guessing. You might be able to make it sound less bad by fiddling with it, but you'll likely not completely reverse the compression process, and even that's assuming that the record company used a linear (as opposed to variable) compression algorithm. Recordings shy in bass or perhaps treble can often be improved by the judicious application of an equalizer, but this works better on the bottom of the audio scale than it does on the top. Boosting highs to restore them runs the risk of introducing noise and perhaps distortion that was inaudible before being boosted. And that brings me to the last caveat. If the recording itself has high amounts of distortion in it, there is nothing you can do to eliminate it.  

[CuteStudio]

19 hours ago, Apesbrain said:

 

Any DAW that supports VST plug-ins should be able do the job.  Audacity and Wavosaur do and both are free. 

 

:

 

but it's easy to pour on too much and get crap.

 

The problem that the record companies give us is that each track is mangled in a different way, even on the same album exactly the same mangling process is rare.

 

So what will work well with one track will be terrible with another, 50% of the problem is analysing what's wrong and where to fix, in fact in some ways the fix is easier than the diagnosisi.

 

The above track from Joss Stone has a 5% difference in the clip positions of each channel and around 2% difference between the +ve going and -ve going parts of the waveform. It has a combination of compression and clipping, in this case the +ve side has a 'wandering' clip which looks like compression but is in fact just untidy clips. 

Also the clips themselves vary, some are dirty (note the tiny overshoot above), some have VHF noise, some have a tiny signal trapped within.

 

Even if you could pick the perfect dynamic reconstruction for this there are two problems left:

 

1) The clips will be louder still (more HF noise)

2) All the peaks are often exactly the same level, from single points resting on the top to long clips. So your expander will have now created a taller brick with some pretty wild spikes formed from elongating the narrow bits, which exceed permissible rise times and create noise.

 

The overall view of that same track. Note bit0 has effectively been removed at mastering time, makes a mockery of that care in the playback system doesn't it.

 

The irony is that the tracks that are easier to improve dynamically are therefore by definition the ones that need it least of all.

EQ is a red herring, but can reduce clip noise.

 

Additionally compression and limiting is multilayers, apart from the data loss (equal peak heights) you cannot know what algorithms/formulas have been applied as the mastering mangler will have twiddled his settings to create the terrible sound he is after.

 

If the RIAA was interesting in music instead of money they would have standardised a maximum tampering or compression standard years ago, and then allowed the makers of car stereos to compress it on playback as required. They standardized the LP EQ standard (although arguably dropped the ball there too at the top end) but today seem to concentrate on enforcement to consume their terrible product instead of enforcing a minimum quality.

 

Which is why today if you want to hear a decent CD you generally have to dig out some old (mastered in the 1980s!) Floyd etc.

BTW. 

[sandyk]

A guy who is a DIY Audio member, wrote a Program some time back that can often do wonders with some of the older CDs from the early stages of the Loudness Wars . ( SeeDeClip Duo Pro.)

It usually can't do too much though with "basket case" clips like shown above,  but it may often make them more tolerable.

It may  even markedly improve the audio of some of the earlier Music Video's , if the video is demuxed, and the restored Audio Muxed back in again.

 

 

[fas42]

30 minutes ago, CuteStudio said:

 

The problem that the record companies give us is that each track is mangled in a different way, even on the same album exactly the same mangling process is rare.

...

 

Which is why today if you want to hear a decent CD you generally have to dig out some old (mastered in the 1980s!) Floyd etc.

BTW. 

 

So you're the man behind SeeDeClip ?!

 

I looked at that program about 6 years ago, when investigating my options for "fixing" tracks. I can't recall what the results were like, using the utility at that time - but ended up doing it "my way", back then. Perhaps I should have another look at the latest version ... ??

 

[Audiophile Neuroscience]

7 hours ago, gmgraves said:

That can be reversed using a com-pander like a dBX, IF (and boy is that a big IF) you know (or could figure out) the compression and release slope that the record company used to compress the album in the first place! Otherwise you're just guessing.

 

I still have a dBX from the 80's I think.

[Audiophile Neuroscience]

27 minutes ago, fas42 said:

SeeDeClip ?!

 

I looked at that program about 6 years ago, when investigating my options for "fixing" tracks. I can't recall what the results were like, using the utility at that time - but ended up doing it "my way", back then. Perhaps I should have another look at the latest version ... ??

 

 I had a look at the website but to be honest found it confusing, the website that is. It appears unlike previous versions that modified tracks for archiving it morphed into a client-server thingy using your browser as a media player and analyzing and modifying tracks on the fly. No idea how but certainly happy to try it. Has anybody tried it recently?

[sandyk]

A couple of Sound Forge 9 screen grabs of "Better Be Home Soon - Best Audiophile Voices 2"

One is the original, and the other is newly created using SeeDeclip Duo Pro.

 

 

 

[miguelito]

On 10/6/2017 at 3:04 AM, fas42 said:

Decompression/Declipping ... yes! The experiments I did showed the potential, and I was merely feeling my way, purely explorative efforts. One pop track was just savagely compressed in the final mixing, and the parameters used in that operation could be deduced quite straightforwardly, from examining the waveform - applying the inverse of those parameters, in editing software that allowed uncompressing, did a nice job - there was no loss, from my POV, in the integrity of the track, but the massaged result sounded very much like a straight mix of the instruments.

 

Beware! There are plenty of utilities and such out there that claim to do things, but my experience of them is not good - I went back to basics, in analysing what could be seen in the waveform, and from that deduced what could be a good approach.

 

What tools have you used for decompression? I fear that compressed music has also a lot of clipping, which would make decompression impossible.

[fas42]

13 minutes ago, sandyk said:

A couple of Sound Forge 9 screen grabs of "Better Be Home Soon - Best Audiophile Voices 2"

One is the original, and the other is newly created using SeeDeclip Duo Pro.

 

 

 

 

Hmmm ... unfortunately, I don't see in that too much undoing of compression effects, more just straight declipping. Which is what I found when I investigated the software back then - the algorithms didn't have the ability to analyse what had been done to the track, and couldn't do sensible reversals.

[miguelito]

8 hours ago, gmgraves said:

For instance, many modern pop recordings are heavily compressed. That can be reversed using a com-pander like a dBX, IF (and boy is that a big IF) you know (or could figure out) the compression and release slope that the record company used to compress the album in the first place! Otherwise you're just guessing.

 

Interesting, though I wonder if compression is really that fine tuned as opposed to using a universal setting. I also wonder if analyzing an entire song can be helpful to estimate compression parameters. There are many instances in data analysis where transformed can be “deconvoluted”. If there is clipping, then you’re screwed as that loses info.