Can Bad Recordings be Fixed?

[semente]

Just now, fas42 said:

 

The answer is vastly more than a simple No, or Yes. First all all, why is the recording "bad" - was the equipment used in terrible shape, of poor quality, and imposed a strong distortion signature on the end result; ot was the recording gear of superb quality, the best one can buy, but the mastering "mangled" the raw inputs? There is a huge spectrum here, and the solutions will, or may vary per recording.

 

I agree that the effectiveness of the "treatment" is inversely proportional to the "disease" (and it's also genre or better programme specific).

That's all.

 

The degradation starts from the mic onward and any editing of the signal will take its toll.

Mixing and EQ'ing included.

[fas42]

Just now, semente said:

 

I agree that the effectiveness of the "treatment" is inversely proportional to the "disease" (and it's also genre or better programme specific).

That's all.

 

The degradation starts from the mic onward and any editing of the signal will take its toll.

Mixing and EQ'ing included.

 

Yes, degrading will always take place - it cannot be otherwise, because of the nature of the universe; unless the source is purely digital in form a perfect replica can't be captured. However, the "great miracle" is, that if all due care is taken, with possible 'correction' of the captured data, and then the absolutely highest standard for the replay setup, that subjectively any flaws that can't be easily eliminated matter not one iota - the listening experience is as full, vital, powerful, emotional, as one could wish for - and isn't that the point? Anyway, such is the case, and the goal, for me.

[CuteStudio]

On 11/23/2017 at 10:22 PM, semente said:

The degradation starts from the mic onward and any editing of the signal will take its toll.

Mixing and EQ'ing included.

 

You are right of course but as fas42 points out; many parts of that chain create very little distortion.

 

I would argue that the vast majority of modern music arrives at the mixing desk in glorious 24bit quality, the problem we have is that when it leaves that desk it's often like this:

 

 

Which is the resultant track we're supposed to pay real money for to love and treasure.

 

Note the odd histogram shape - this album has that all the way through, it's not so unusual to see slight differences between the +ve and -ve signals but it's almost like they used a different limiter for each one here. 

 

I think this was done by an analog limiter, which means an extra DA - interconnects - limiter - interconnects - DA stage for those HiFi purists to consider.

 

So somewhere sitting on a HDD or tape in a locked room in some faceless building is an unclipped, unlimited, uncompressed perfect copy of that album - perhaps as multi-tracks too - but what we have access to is a poor facsimile. It always seems like the artists go to a lot of effort to get the tune, the lyrics, the sound that they way and pay tens of thousands for a good studio, then when they walk out the door assuming their job is done the 'industry' moves in and proceeds to sell the public a copy that for many is unlistenable; and certainly not a track you'd want to listen to at a decent volume.

 

I suspect not many artists go into a record shop, buy their CD and then analyse the mastering quality - the odd artist (e.g Katie Melua) does seem to have more than their fair share of well mastered CDs so maybe she does check what goes out of the door to catch any howlers like the one above.

 

[sandyk]

Hi Graham

 Many thanks for your indepth posts.

Regards

Alex

[semente]

There's hardly any mention of producers in these conversations but I think that they have at least as much responsibility as the performer with the quality of their products.

[sandyk]

1 minute ago, semente said:

There's hardly any mention of producers in these conversations but I think that they have at least as much responsibility as the performer with the quality of their products.

 

 I think that you may find that the guys in the suits ( execs) tell the Recording and Mastering Engineers what they expect from them. (loud and louder!)

[semente]

3 hours ago, CuteStudio said:

 

You are right of course but as fas42 points out; many parts of that chain create very little distortion.

 

I would argue that the vast majority of modern music arrives at the mixing desk in glorious 24bit quality, the problem we have is that when it leaves that desk it's often like this:

 

 

Which is the resultant track we're supposed to pay real money for to love and treasure.

 

Note the odd histogram shape - this album has that all the way through, it's not so unusual to see slight differences between the +ve and -ve signals but it's almost like they used a different limiter for each one here. 

 

I think this was done by an analog limiter, which means an extra DA - interconnects - limiter - interconnects - DA stage for those HiFi purists to consider.

 

So somewhere sitting on a HDD or tape in a locked room in some faceless building is an unclipped, unlimited, uncompressed perfect copy of that album - perhaps as multi-tracks too - but what we have access to is a poor facsimile. It always seems like the artists go to a lot of effort to get the tune, the lyrics, the sound that they way and pay tens of thousands for a good studio, then when they walk out the door assuming their job is done the 'industry' moves in and proceeds to sell the public a copy that for many is unlistenable; and certainly not a track you'd want to listen to at a decent volume.

 

I suspect not many artists go into a record shop, buy their CD and then analyse the mastering quality - the odd artist (e.g Katie Melua) does seem to have more than their fair share of well mastered CDs so maybe she does check what goes out of the door to catch any howlers like the one above.

 

 

Fortunately for me my main staple is Classical.

[gmgraves]

On 11/23/2017 at 2:34 PM, fas42 said:

 

Yes, degrading will always take place - it cannot be otherwise, because of the nature of the universe; unless the source is purely digital in form a perfect replica can't be captured. However, the "great miracle" is, that if all due care is taken, with possible 'correction' of the captured data, and then the absolutely highest standard for the replay setup, that subjectively any flaws that can't be easily eliminated matter not one iota - the listening experience is as full, vital, powerful, emotional, as one could wish for - and isn't that the point? Anyway, such is the case, and the goal, for me.

 

You'd think they could get it right occasionally, if only by accident, but apparently not. I have never heard a commercial recording that sounds as good as it should sound, or even as good as it could sound! Truth to tell, record companies don't make recordings for audiophiles (even when they advertise that they do), they make recordings that can be played on the lowest common denominator of equipment. Case in point. I was at Paul Stubblebine's studio in San Francisco a number of years ago listening to him working on the master tape of the Stanisław Skrowaczewski/Minnesota Orchestra recording of "Ravel's Complete Works for Orchestra", that was originally recorded by Joanna Nickrenz and Marc Aubort in 4-channel sound for release as a Vox Box in QS "quadraphonic" record set back in the 70's. The performances were great, and it was a landmark recording project except that Vox's pressing quality was so poor, that all the sets made were plagued with underfill, re-grind, warped and eccentric records, that it took transfer to CD in the late eighties to be even playable.

Anyway, Stubblebine was remixing the master tape of the "Daphnis et Chloe" for Mobile Fidelity who were going to release it on their audiophile series CDs. It sounded tremendous, in fact, it was jaw dropping through the studio's monitor speakers (which I recall were the large B&W floor-standers, but I could be mis-remembering here). When the CD was released, some months later, I had Mobile Fidelity send me one. What a let-down! It sounded nothing like what I heard in Paul Stubblebine's studio! It was thin, and glassy sounding without much dynamic range. The next tim I ran into Stubblebine, I asked why the released record sounded so poor when compared to the master (which was analog tape, by the way). Paul said that even though it was advertised as an audiophile edition, the reality was that no record company can afford to make recordings just for audiophiles. They can give a recording perhaps a bit more bass, and a bit more dynamic range, but nowhere like what even a RedBook CD is capable. People who weren't audiophiles would scream bloody murder at such a release and most music critics would pan it for requiring the listener to continuously turn the volume up and down to listen to it on their cheap players. Indeed, when compared to the earlier Vox CD release of the same material, there was little difference between them.

 

I don't particularly like pop and rock, Never have, but there are exceptions. I first heard Santana's "Abraxes" recording as a playback in the Wally Heider recoding studio in San Francisco in the early '70's. I was quite taken aback by "Oye Como Va" through Heider's huge JBL monitors in their main control room, The cleanliness of the electronic organ and the purity of Clarlos' electric guitar (without a fuzz box or wa-wa, or any of the other things that make electric guitars my version of the worse sound on earth!) really impressed me. After that, I never minded hearing that piece or Black-Magic Woman. Recently, seeing that a 24/96 version of "Abraxes" was available on HDTracks, I decided to spring for it, thinking that I'd finally have access to a copy of the master that sounded at least as good as the playback I heard at Heider's studio back in the day. Boy was I gypped! the "high-res" version sounded exactly like the recent RedBook CD of the album that a friend brought over so that we could compare - lousy, both of them! (if you haven't heard the master - even through a pair of TERRIBLE JBL studio monitors*, you probably don't know how good the master tape was!).

 

*The recording industry standardized on JBL's back in the 1960s as studio monitors so a tape recorded in, say, Nashville, would sound the same when played back in say, NYC or LA. It was like choosing the sewers of Paris as the standard for how all municipal buildings should smell all over the world. Yeah, they would all smell alike but why would anyone want them to all smell like that???!!!

[fas42]

3 hours ago, gmgraves said:

I don't particularly like pop and rock, Never have, but there are exceptions. I first heard Santana's "Abraxes" recording as a playback in the Wally Heider recoding studio in San Francisco in the early '70's. I was quite taken aback by "Oye Como Va" through Heider's huge JBL monitors in their main control room, The cleanliness of the electronic organ and the purity of Clarlos' electric guitar (without a fuzz box or wa-wa, or any of the other things that make electric guitars my version of the worse sound on earth!) really impressed me. After that, I never minded hearing that piece or Black-Magic Woman. Recently, seeing that a 24/96 version of "Abraxes" was available on HDTracks, I decided to spring for it, thinking that I'd finally have access to a copy of the master that sounded at least as good as the playback I heard at Heider's studio back in the day. Boy was I gypped! the "high-res" version sounded exactly like the recent RedBook CD of the album that a friend brought over so that we could compare - lousy, both of them! (if you haven't heard the master - even through a pair of TERRIBLE JBL studio monitors*, you probably don't know how good the master tape was!).
 

 

Sometimes it's the surprising recordings that get it together - I've mentioned elsewhere an album by Alligator Records, Hound Dog Taylor - A Tribute ... house rockin' blues. The dynamic bite and life of this is something to behold - I haven't tried analysing the waveform, but subjectively it feels like it's mighty close to the raw feeds of the microphones, and nothing else.

[fas42]

12 hours ago, CuteStudio said:

 

You are right of course but as fas42 points out; many parts of that chain create very little distortion.

 

I would argue that the vast majority of modern music arrives at the mixing desk in glorious 24bit quality, the problem we have is that when it leaves that desk it's often like this:

 

 

Which is the resultant track we're supposed to pay real money for to love and treasure.

 

 

 

I'm confused at the moment - I've got hold of the audio, of that song, "You've Ruined Me" - and the waveform looks nothing like the above - remarkably good dynamics, just by the visuals. Am I looking at the "wrong" version?

[CuteStudio]

On 11/25/2017 at 11:16 PM, fas42 said:

 

I'm confused at the moment - I've got hold of the audio, of that song, "You've Ruined Me" - and the waveform looks nothing like the above - remarkably good dynamics, just by the visuals. Am I looking at the "wrong" version?

 

Possibly not, I may be looking at a different version, I'm going to dig out my original CD from the loft this week (when I can brave the cold and finish with the 10+ CDs currently littering the office!!) so hold on until I confirm that I was actually playing what was on that disc and not some hacky mp3 (I'm still re-ripping stuff to WAV - I used to need an mp3 copy of stuff for my iPod).

[CuteStudio]

Here's a better example of uneven limiting I found from Kate Earl's 'Fate is The Hunter' album. Overall a brilliant, 'must buy' CD and about half the tracks are not badly compressed or limited at all, which is rare. The track 'Officer' is a lot better than most modern stuff, but you can see it's still flawed.

 

What you should see in audacity

 

Analysis

 

 

Below is an more typical example of CD limiting, it's very compressed and also limited, the spike at the end of the histogram (RHS of the top-left) is due to mild clipping (the spike if formed by counting the samples sitting at the edge, there's a concentration of them due to the clips).

 

 

The high number of samples and the sharp edge there denotes some major limiting so in practice all the peaks are rough flat-tops, although the musical content of this track doesn't create many very long clips, the longest being 58 samples and the average just over 6 samples.

 

Here's a control track - Dire Straits, Money for Nothing, showing what a natural histogram should look like. Note the sparsity and natural taper. This actually does have a few points clipped here and there so they're using the full CD and it does sound magnificent. It sounds good in the car too, even a 1980s car to match the time  

 

 

Note here the thinness of the bright RMS area, compare it to the fatter RMS area shown in the Kate Earl and Gainsbourg tracks. Those tracks have a far higher RMS level than the older ones, but basically the same peak level indicating not just the higher amount of compression used today, but also the limiting and clipping.

 

It's important to discern the difference, flattened peaks are caused by limiting and clipping, compression itself has a more subtle affect on fidelity.

These two rather separate mechanisms help prevent us from expanding out the damage and most of the harshness comes I think from the squarish shapes of the peaks generating IM distortion.

[fas42]

The Officer track has something I hadn't seen before - the waveform overall has a good look to it, but there is a clocklike beating element which is compressed throughout the track. Which is a giveaway; it's the steady drumbeat, every 0.6 secs, which has been knobbled - again, I attenuated by 1dB to eliminate the reconstructed waveform clipping.

 

I'll have a go at undoing that beat compression, see if I can get it in better shape.

[fas42]

Okay, got somewhere - in Audacity, original attenuated by 3dB, and that expanded to get rid of the ruler edges to the waveform. Playing back, no obvious artifacts; and a lot more bite from the backing instruments; no change to the voice.

 

 

[CuteStudio]

On 11/27/2017 at 11:34 PM, fas42 said:

Okay, got somewhere - in Audacity, original attenuated by 3dB, and that expanded to get rid of the ruler edges to the waveform. Playing back, no obvious artifacts; and a lot more bite from the backing instruments; no change to the voice.

 

 

 

Interesting, it always amazes me how important those little peaks are, more bite or 'snap' from the backing is a good description, for me it's a little more like the instruments are in the room.

 

BTW I did get around to re-ripping Norah Jone's 'The Fall' into WAV and it isn't actually any worse than her other albums now. I then lost interest in that (although I liked the tunes) and decided to re-rip a few more including Solace by Sarah McLachlan and that was much better too, so at some stage I must have ripped at least 2 of my CDs with an MP3 encoder that did more than just encode them. Perhaps an early iTunes or Lame, not sure.

 

Sadly when I tried re-ripping Katie Herzig's 'Waking Sleep' it was indeed just as bad, but at least it cleans up well to sound quite good in the end. I don't think I've got any left to re-rip now but at least 2 albums have much better sources now, which is a result!

 

Herzig's chop:

 

 

 

[Mike27]

I wish I had run across this thread 4 years ago. I was just getting started using “Perfect Declipper.” Life being full of distractions, I didn’t get to focus on it until a couple years ago when I found some time. Now I have about 12,000 declipped tracks & still plugging away at it. It’s surprising the number of CDs that are clipped. ‘50s jazz, all manner of stuff. Here’s “Return to Sender” from 2007’s “The Essential Elvis Presley.” Top stereo pair as ripped, 2nd pair reduced 6.02 dB for comparison with the third, declipped tracks. 
 

We’ll likely never know what was lost in the Universal fire.