About noise and de-noizing... from Eroc

I got this personal original (english) message from Eroc.

What kind of noise in a recording can be called "authentic" or "original"...?
The noise from the studio gear - says the engineer. The hiss from the
tape-material - says the masterer. The noise from the vinyl material - says the
customer. The (low) noise from the home equipment - says the HiFi-freak. And:
the noise from the my wife with the vacuum cleaner downstairs - says I...

But no artist and no producer ever in the whole world attended to put more noise
and hiss into his recordings "as necessary" because of lack of the recording and
reproducing technology. And each musician and producer in this whole world (and
I speak for nearly them all, because I know them nearly all) always was very
delighted during the studio sessions, when the engineer offered him some methods
to avoid hiss and noise in his recordings.

So - as a remasterer I fanatically try to get rid of each little bit of hiss and
noise, which occured due to technical resons in the recording process. But I
hardly try to keep each little bit of hiss and noise which is "authentic", e.g.
the typical "cooking" of an AC-30 oder the growling breath of a Leslie cabinet,
or even the slight rumbling of the traffic outside the Vienna opera during a
violine solo.

Boy, I know what I'm talking about. Just finished remastering 240 titles for
Line-Music from the 40's. And they all were my guests: Benny Goodman, Stan
Kenton, Peggy Lee, Jo Stafford, Louie Jordan, Les Paul, The Weavers, Doris Day,
Frank Sinatra, Vaughn Monroe, Ella Fitzgerald, Rosemary Clooney, Ray Anthony,
Hank Williams sr., Lloyd Price, Guy Mitchell, Count Basie, The King Cole Trio,
Kay Starr, Gene Krupa, Jimmy Dorsey, Tennessee Ernie Ford, Johnnie Ray, Wynonie
Harris, Glenn Miller, Bing Crosby, The Ink Spots, Tex Ritter, Frankie Laine, Guy
Lombardo, Les Brown, Artie Shaw, Dean Martin, Cab Calloway, The Andrew Sisiters,
John Lee Hooker, Duke Ellington, Dinah Shore, Judy Garland and many many more.

And each one of them was sitting here besides me with a glass of champaign and
sparkling eyes, not believing what modern mastering technology can whip out of
these old dusty recordings of what we can call probably the best music ever
made. And the only thing I could do to it with my modest knowledge and my
"musicians heart" was, to make that music come over from the humans who made it,
to us humans today in the best possible way.

I succeeded. Never before all these hits could be heard and "adapted" in that
way. Most of the folks around here don't believe that these are just the
original recordings. And when it will be released soon, especially the Americans
will be in first row to not believe it, too. Until they have tried to make it
better (as always).

Right on, fellows...


- Eroc

Image from Eroc

 

Tape hiss is the hardest thing to get rid of without either introducing artifacts or changing the sound.

 

Just a guess here, but it will probably take far higher bit rates than we have now, including DSD, to apporach the ability to remove tape hiss. Not to mention the procesing power it will require to do it in the digital realm.

 

The bit-depth isn't so much the problem as is the algorithm's ability to tell the hiss from actual music, although working at higher bit-depths help. Don't know if higher sampling rates make a difference, though...

 

I agree. I just wonder, and I know enough only to be dangerous , if algorithms would require much higher bit rates to distinguish "noise" from part of the music. If they could create a tool in digital editing, kins like a "magic wand" that selects a type of noise, then examines the track for all areas that match that noise, then remove it, leaving every thing else as is. not just a tape his plug in or a ground loop hum remover.

I can only dream

 

Some investigation on this man came up with this:

Eroc

 

Amen!

What bothers me is this infatuation people have for removing hiss, distortion etc.

Firstly, you can't "improve" the quality of an analogue recording by processing it digitally. The original analogue tape is, by definition, the best version of the recording there is. By loading it into pro tools and manipulating the digital copy you destroy the integrity of the original recording (If you've ever been dissapointed with a Peter Mew CD you know what I'm talking about!).

Secondly, The hiss, distortion etc.. is part of the original recording. As a kid, I was fascinated by the sound that came from recordings, warts and all. For me, the pleasure of listening to music is the magical sound of the recording in all it's glory - including any background noise inherent in the recording process. Anyway, it's not as if the hiss is overbearing on good recordings from decades past. Removing the hiss sucks the life out of the recording, surely it is sensible to leave it there.

Just because we live in an age where we have the option of digital recording (and therefore no tape hiss) doesn't mean we all want to go back and remove the tape hiss from old analogue recordings.




Digital Noise Reduction is KILLING music!!!

 

As much as I disagree with Eroc's theories, the original analog tape is *not* necessarily "the best version of the recording there is." As I've mentioned before, if it was, Steve would be out of a job - there'd be no reason to master anything.

I doubt that many people would prefer the sound of the Mamas & Papas master tapes to Steve's CD, for example.

ProTools isn't the devil. You can do some great (sounding) stuff in PT. It's specific things like NR and digital compression that are the evils...

 

This isn't a black and white issue folks -- an untouched master vs. a "bad" mastering engineer. Luke speaks the truth.

We are all being backseat masterers making assumptions that only Steve and other mastering engineers can verify, but the truth is in the listening of the final product. I don't think there are many masters out there that we'd want to hear before a good mastering engineer got through with them. A good sounding disc that didn't require changes from the flat master is probably the exception not the rule.

I think Steve said recently that RINGO was one of the easiest times he's had because the master sounded so good flat -- that sounds like an exception to me.


- Gabe

 

Well, what I'm saying is that the best analogue tape source at your disposal can not be improved by digital processing.

That is not to say you can't make adjustments to the equalisation to obtain a more preferable tonal balance, particularly in the case where there is a significant problem with the original equalisation such as with the Mamas & Papas tapes.

Steves does all his mastering tweaks in the analogue domain thus maintaining the integrity of the original recordings......

When transferring to digital the goal, in my eyes, is to sample the analogue waveform and record the digital qauntisation of each sample - and that's it!

Any further manipulation of the digital data produced from the sampling process that changes that data DESTROYS the integrity of the original recording. This is the key to making good sounding digital recordings - do not, under any circumstances, manipulate the data after the sampling process.

Why would you expect the recording to sound natural after digitally manipulating the sampled data - it wont.

 

Eroc looks like the type that has mixing boards that go to 11.

Tsk, tsk, tsk... BAD Eroc. Bad boy.

 

So you honestly believe that *any* type of digital processing will "destroy the integrity of the original recording?" While one can certainly prefer analog processing, I think that's a really bold and unfair statement to make. Are you saying you only like CDs that were mastered in the analog domain? If so, I guess that means you hate most CDs out there today, since the majority were mastered digitally.

How about albums that were recorded and mixed digitally? Do those albums have no "integrity"?

A better question (in my mind) would be why *wouldn't* you expect the recording to still sound natural? What is so inherently "unnatural" about digital processing?

 

Got it in one

Well, I was saying that the integrity of an analogue recording is destroyed if you sample it and then process the data but the same applies to a digital recording. If you were to record acoustic instruments digitally, then as long as you don't then process the data the integrity is maintained. As soon as you change the samples you've broken the golden rule.

The accurate reproduction of acoustic sounds is a fine art. Sampling an analogue waveform to produce a digital representation of it is fine (in PCM the higher the sampling frequency and the longer the wordlength the better - in DSD the higher the sampling rate the better) but once in the digital domain it is paramount that you don't alter the samples if you want to recreate the original analogue waveform as closely as possible on playback. That's just common sense!

Of course, there are different levels of processing and the further you get from the original data the worse the results - NoNoise would be an example of heavy processing.

It's no coincidence that all the CDs in my collection that sound the most natural and that I enjoy listening to the most are the ones that adhere to the "golden rule" most closely .

 

The only difference between an analog recording and a digital one is the digital one didn't get recorded to (analog) magnetic tape first. The signals being recorded are analog in both cases.

The *only* way you wouldn't process things in a digital recording is if it was done straight to stereo, with *no* additional mastering. *Every* multitrack digital recording mixed through a digital desk has processing. Volume adjustments are processing. Panning is processing. Heck, combining multiple tracks is processing. Let's not forget EQ.

Do you honestly believe that every single multitrack digital recording is "bad" or has no "integrity"?

If your only goal was to recreate the "original analogue waveform as closely as possible", you wouldn't be doing any processing in the first place, be it in digital *or* analog. That's what *any* type of processing does - it *changes* the waveform. EQ *changes* the waveform.

Why is it "common sense"? What does it matter if you change the waveform before it is digitally sampled or after?

But how do you know that? Have you confirmed with the engineer on every CD you feel sounds the most natural that *no* processing was done digitally?

 

Yep!

As I said - sampling an analogue source (be it from a tape source or live or whatever) and recording the quantisation words for each of those samples (in the PCM case) is one thing but then changing those samples (ie digitally prcessing them - in any way whatsoever) is quite another. What you end up with is, fundamentally, no longer a representation of the original waveform. The whole basis on which digital represenration on an analogue wave can succeed relies on you not changing the samples once you have calculated them!

In mastering an analogue recording, you may desire a little adjustment to the tonal balance of the signal. Doing this in the analogue domain does not destroy the integrity of the original recording - sure it changes the waveform - but it still is a continous analogue waveform and will sound like one. If you perform the EQ in the digital domain you end up with a digital representation of a waveform that has never existed in the analogue domain. When you convert this digitally created waveform back to analogue it doesen't sound natural to me.

This is the key point!! If you change it in the analogue domain then at least you still have a real, continous analogue waveform. Convert that to digital, and as long as you don't process the data, you can accurately recreate it on playback. Process it in the digital domain and on playback you are reproducing an analogue waveform that never existed in analogue before - do you think it is common sense to assume that it will sound OK?

RCA Bowie, DCC golds, Japanese "Abbey Road" etc, etc......

You don't need to be Einstein to realise that the most natural, analogue sounding CDs (I'm talking about re-issues of analogue recordings here) are the ones with the least digital processing (ideally none!).

 

NOT if one is careful and knows what they are doing. Aside from NR, there are a multitude of things you can do to music and still not lose it's integrity. I think some of you are getting the idea that all Pro Tools is just NR. In fact, NR and compression are NOT the devil. They are just tools. The problem is simply that some engineers tend to OVERDO things, and recklessly!

Digital isn't bad. It's just another way to work with music. The results depend on the skills of the user.

 

Yeah, that's what I've been saying...

 

Malc, if you only knew what went on in the recording/mastering of nost CDs, and even many LPs of today, you would withdraw your claim.

 

It's no less a representation of the original waveform than if you were to process them in analog the same way.

Of course it hasn't - you're *trying* to change the waveform.

You seem to be saying that digital processing must be "wrong" because the waveform produced never existed in analog. Well, let's take it a step further. *Any* processing is "wrong" because the waveform produced by it never existed in the original signal.

How about the case where one both EQ's in analog and transfers to digital, and transfers to digital and EQ's in digital. And the end waveform comes out to be the same. What does it matter *how* one got there? How is the one processed digitally any less "accurate"?

What are you basing that off of? Have you heard every different type of digital processing equipment available? Have you heard every type of digital to analog converter available?

It's one thing to say "I don't like what I've heard." It's quite another to say "digital processing must be inherently flawed since I don't like what I've heard."

Please, see above. Output from a digital source *is* a continuous analog waveform.

And have you confirmed in every case that *no* digital processing was done?

 

Now we're speaking with three heads!

Luke and I have been at odds on all kinds of things musical! But, we both have a lot of experience working with digital.

The thing is that digital is just another tool. I think a lot of hatred or mistrust of digital comes from a lack of knowledge about how digital works, and negative experiecnce with the worst of it.

 

That's funny!

 

What he don't know!

 

NR is one of the most extreme examples, but any digital processing that changes the sample values is going to compromise the integrity of the original waveform.

The Nyquist theorem states that by sampling an analogue waveform you can faithfuly reproduce the analogue waveform in the frequency range from zero up to half the sampling frequency. Obviously, the higher the sampling frequency and the longer the wordlength (in PCM) the more accurately the analogue waveform is reproduced.

However, if you go around changing the values of the samples, you have DESTROYED the integrity of the original signal in that when you convert the samples back to analogue you no longer have the original waveform.

Sampling enables you to encode an analogue waveform digitally.

ProTools takes those samples and messes around with them to do things to the waveform - this is not compatible with accurate reproduction of analogue recordings.