Demastering your music

[Chris A]

Demastering your CDs from the effects of mastering equalization, limiting (clipping), and residual line noise (50/60 Hz, 100/120 Hz) and HVAC infrasonic noise to improve the sound is described below. The story of why and the process described using an audio editing tool, such as the freeware tool Audacity, is part of a multi-part tutorial.

The first segment of this tutorial is the "what", "why", "who" and "where" of the music mastering problem (see pdf enclosure).

Part 2 including the "how", "how much", and "when" is also available, below

Part 3 (advanced topics) will follow shortly.

Chris

 

[Peter Loeser]

Very interesting topic Chris - thanks for contributing. I've used Audacity in the past but not quite to this extent. I'll have to take some time to read through the tutorial, but I'm definitely interested in giving this a try. Looking forward to hearing how it works for others too.

 

[Todd Anderson]

Very interesting topic... frankly, not something I've ever considering exploring. When I have a moment this week I'll give these PDFs a read. Glad you're starting this, Chris!

 

[Tony V.]

So in some ways to simplify this your applying a "house curve" to the master so it plays more like it should with more fullness particularly on the bottom end? Loudness could also be the concept here but rather applying it directly to the ripped recording.

I agree with this idea and may give it a go to some older CDs that simply sound flat with no dynamics to them.

 

[Chris A]

So to provide a little more background to the subject, there is a wide divide between actual as-recorded fidelity and that those stereo tracks which are typically played in forum members' listening rooms from CD or even worse, phonograph records.

One notable author that has talked at some length about this subject is Philip Newell in his book Recording Studio Design:

"Much popular music...never exists as a single integrated performance in any one time and place until the final mix is first heard in the control room...The concept of 'real' reference is therefore lost...a recorded bass drum, for example, rarely sounds like the actual drum [as discussed in part 1 of the tutorial above]. The recording of bass drums has now become highly stylised."...

...there are some studios which are liked for their inspirational capacity during the recordings, but which during mixing would lead to unrepresentative results...but such idiosyncratic studios must be considered to be more of a performance stage (control room included) than a general reference for what is 'right' about a balanced mix...

...Somewhat perversely, we have returned to a situation whereby the vast majority of control rooms are once again not properly designed, so it is little wonder that close field monitoring is still in widespread use...Listening close to or within the critical distance also helps to remove a degree of room-to-room variability. Unfortunately, though, this does not tell the whole story about the recording...[the effects of close monitoring] can be deemed totally undesirable when the music is heard on a full range, truly high fidelity domestic system.

...It does not seem professional to deliver products [i.e., audio recordings] that are exposed as lacking by the purchasers of audiophile systems. In other words, the quality heard from a recording should be proportionate to the quality heard from the quality of the system on which it is played. All too often, due to inadequate monitoring systems in the studios, the audiophiles in their homes are the first to realise just how awful some recordings are."​

 

[Chris A]

So in some ways to simplify this your applying a "house curve" to the master so it plays more like it should with more fullness particularly on the bottom end? ...I agree with this idea and may give it a go to some older CDs that simply sound flat with no dynamics to them.

Well, what the technique is doing is reverse engineering the mixing+mastering equalization curves, and applying that reverse equalization curve. When you hear the results, I think that you'll agree.

The worst CDs in terms of mastering EQ applied I've found (but with intact dynamic range) is in CDs from the mid-late 1980s, up through 1990. In 1991, multi-band compressors began to come into widespread use, and the techniques shifted away from strictly EQ to make the mix louder. At that time, the multi-band compressors combined with limiting ("clipping") to make the finished product louder.

However, some of the worst EQ that I've seen thus far is on a Smashing Pumpkins CD Siamese Dream (1993). Here is the net mastering EQ curve for the track Cherub Rock (i.e., the inverse of the demastering curve):

 

[Chris A]

If anyone wants to hear some of the results of this technique applied to commonly owned CDs, just send me a personal message.

Chris

 

[Todd Anderson]

I'd love to hear the differences... I have to admit, I have quite a few CDs from that era that just don't sound great to the ears. I've always found it very disappointing, but never thought much beyond the fact that it must just be due to what technology was available at the time!

Curious, roughly how long does it take to clean up a song?

 

[Chris A]

Well, the time per track varies based on instrumentation and the degree of "mastering EQ" used on each track. It also takes me far less time now to demaster a track than it did when I started 2 1/2 years ago.

As an example, I recently demastered two Allman Brothers CDs (Enlightened Rogues and Brothers and Sisters). As is the case with popular music CDs, every track had different/unique EQ curves applied (as opposed to the same EQ curve used on many classical albums). I was able to demaster and save each track before the preceding track completed playing as I was creating the reverse-engineered EQ curve for the next track, applying it to the track, checking its final response in "plot spectrum" and "Spectrogram log(f)" views and adjusting the EQ, then reapplying the updated EQ curve, then saving. So I demastered both CDs (all having different EQ correction curves for each track) in less time than it takes the CDs to play.

Sometimes the demastering is more involved due to the instrumentation in the tracks (such as solo voice, piano, violin, etc.), the degree of mastering EQ used on each track, how different each reverse-engineered EQ curve is from track to track, and the sensitivity of the mix to fine-grain adjustments. I've had some CDs require 10-15 minutes per track or more under those conditions.

The easiest CDs to demaster are those with full instrumentation (like a rock band), tracks that are fairly short (less than 10 minutes duration per track), and the mastering person reused his mastering EQ curve with variations from track to track.

The most difficult CDs are usually solo classical instruments or voices (a capella) with thin instrumentation and a heavy-handed mastering person.

Chris

 

[Sonnie]

Very Interesting!

 

[Sonnie]

I have often wondered about James Newton Howard and Friends.... sub bass has always been lacking on that CD, but I just figured there wasn't any.

Is the MFSL version a culprit too? It sounds the same... lacking sub bass.

 

[Sonnie]

Hmmm... I got lost on the demastering. I actually have Audacity on my computers... use it every week to edit recorded lessons, but I am not following the demastering instructions too well.

 

[Chris A]

James Newton Howard & Friends has fairly mild mastering EQ applied to it, but for instance, the demastering curve for the first track, Caesar:

This is fairly typical of the other tracks on that album, although I found each track's demastering EQ to be unique.

As far as the demastering instructions, I'll take a look at decluttering the descriptions and filling in any gaps. The demastering process is in those basic steps, but the steps could be clearer. Those basic steps were written down 2 1/2 years ago...

Chris

 

[Chris A]

One of the things that I've found about "remastered" CDs like MFSL and even produced from the original corporations that own the music, is that they typically aren't "demastered", but rather compressed more and limited (clipped) more to make them louder. It seems that most remastering jobs actually do not touch the basic mastering EQ curves that were originally applied, however poor and unnatural sounding those curves might be.

The demastering process described here actually relaxes that applied mastering EQ. The harmonics of all instruments and voices begin to sound much more natural and "acoustic".

In fact, once the demastering EQ starts to get close to the finished demastering EQ curve, the sensitivity of the overall track to extremely small changes typically become pronounced, as if the track is rebounding back to its original shape and is again responding to very fine grain changes in its shape, requiring perhaps 1-2 dB changes to taste--instead of the typical 6-12 dB changes (or more) impressed by the original mixing/mastering people.

Chris

 

[Sonnie]

What are you using for a mic to listen to the music via Audacity?

 

[Chris A]

I'm not using a microphone with Audacity, rather I'm looking at the PCM music track ripped to external HD...FLACs from CD/hybrid SACD/DVD-A/download PCM files. (There are no lossy tracks in my music library.)

The calibrated microphone comes in when I flatten the setup's in-room response using REW and an active crossover (Xilica XP-8080). The setup is presently dialed in to ±2 dB from 33 Hz to 17 kHz at one metre. That's what I'm using to listen to the tracks as I fine tune their demastering curves.

Chris

 

[Sonnie]

Okay... so the instructions say to play back the track thru your flattened frequency response speakers. What is the purpose of this if we are not trying to view the music via Audacity? This is what seems confusing to me.

 

[Chris A]

Audacity is the tool we're using to demaster, play back for auditioning, and save music tracks.

Before demastering can commence, however, our setups must have very flat frequency response--approaching a mastering studio, if possible. That's pretty straightforward and inexpensive using REW, calibration microphone, loudspeaker processor/active crossover, and perhaps an AVR/AVP to balance channels and set delays. That capability basically wasn't available cheaply 15-20 years ago. Once you've got your setup frequency response dialed in, then no more use of REW is required. It's all Audacity from that point on.

 

[Sonnie]

What does having my system dialed in have to do with using Audacity? That is what is confusing. Your instructions are as if the system being dialed in is a prerequisite to using Audacity, but the two are not connected. I would take it for granted and assume folks are going to have their systems dialed in as they prefer... no need to tell them that. I am extremely familiar with REW, setup, etc... been using it since it's inception... the author is my former partner. What I want to know is how to use Audacity to demaster the track. If Audacity is not connected to my system, then my system has nothing to do with what will happen with Audacity.

 

[Chris A]

Okay, I think one or two flow diagrams (i.e., things to do before starting Audacity, and how to do it once everything is set up) would clear up the issues that you raise. I'll insert those into the next version of the Part 2 (How To) PDF file.

Thanks for the feedback.

Chris

 

[Sonnie]

We should be able to add these to our Downloads page too. I think a lot of people would be interested in doing this.

 

[Chris A]

By the way...and this is the reason why I talked about using REW in the first place:

The reason for using REW on your setup before starting Audacity demastering is that if you ever want to put your demastered tracks on another system, and if you've demastered using a setup that isn't calibrated flat, you'll find that you'll wind up only playing those demastered tracks on that particular system on which you demastered them. You may not see this point now, but I assure you that you'll see it later, i.e., this is from lessons learned.

Additionally, I've found that I've had little changes in my setup as time passes. Some intentional...many not...and there are lots of little dials on the front of the rack, if you catch my drift. Without returning the system to a known reference point--and this becomes much more important as you restore attenuated frequencies below 100 Hz (more on this subject later)--you're going to wind up having to demaster the same tracks again and again, which is probably something that you'll want to avoid. If you demaster some tracks, then stop for a week or a month, then restart, I recommend a quick recheck using REW before you proceed.

Flat (no "house curve") is recommended during demastering...unless your front left and right loudspeakers have significantly changing directionality vs. frequency, and/or issues with nearfield reflections. Then you've got a choice to make--whether to demaster for that setup only or not. You can add back any house curves that may use at playback time--after demastering. (JMTC.)

I don't use house curves on my setup because its directionality and nearfield reflections are highly controlled. This is really why people are using house curves, in addition to partially trying to cancel the poor mastering EQ that's already there on their music. Since you're effectively trying to remove that mastering EQ using this process, you don't want to reapply it again at the same time.

The visual aspect of using Audacity along with the combined auditioning of the tracks really helps to keep you out of the ditch if something goes wrong somewhere. There is personal time invested in this and making sure that work isn't later found to be negated because of some setup issue is something that's save a lot of work for me. I've demastered over 10,000 music tracks presently--and this is a real issue. I've saved myself many days of work because of attention to this.

Chris

 

[Chris A]

Okay, the subject of restoring bass below 100 Hz.

Floyd Toole talked about this issue recently in a YouTube video taken in Canada. What he said was that, because the human hearing system become less and less sensitive at lower frequencies, very small changes in relative bass SPL below 100 Hz leads to disproportionate differences in perceived bass loudness levels. For review purposes, refer to the following set of equal loudness contours for humans in a controlled environment:

Notice how close together the equal loudness curves are below 100 Hz. Very small changes in loudness measured in dB--once you get the loudness up to audible levels--leads to big changes in perceived loudness as measured in phons or sones, i.e., relative to the equal loudness curves.

If you're restoring severely deficient bass in music tracks, if you're off by just a little on the high side (implying that your setup was just a little bit low in its bass frequency response when auditioning the tracks), you're going to hear it disproportionately as low frequency heaviness or booming when calibrated flat (on another system or your demastering system once you recalibrate it) than if you're demastering in the midrange or lower treble. This is a strong reason to keep your setup calibrated flat--especially below 100 Hz.

Chris

 

[Sonnie]

The reason for using REW on your setup before starting Audacity demastering is that if you ever want to put your demastered tracks on another system, and if you've demastered using a setup that isn't calibrated flat, you'll find that you'll wind up only playing those demastered tracks on that particular system on which you demastered them. You may not see this point now, but I assure you that you'll see it later, i.e., this is from lessons learned.

Okay... so now I a totally confused.

If our system that we demaster on has to be flat to do demastering, then it is apparent that Audacity has to be attached to that system in some way. You keep connecting the system we are listening on to the demastering "process". How are the two connected in the actual demastering process? Previously I understood these to be two separate items.

What am I missing here?

 

[Leonard Caillouet]

Apparently he is talking about using the system as a reference to get a particular "demastering" of a track to sound the way you want it to so if you apply a house curve you have to consider the relative effect on another system that you might be using that track on in the future. The problem I see with this is that someone using a house curve will have to decide whether to use it or not on subsequent listening, because you would presumably be "demastering" to a similar sound that you have chosen for a house curve.

This is all interesting, at the level of a better understanding of mastering practices, but there are far too many variables here that we don't know to consider this "demastering" IMO. What we are doing is simply equalizing to taste or to some arbitrary standard that may or may not be appropriate for other systems. If a particular track or album is really bad, maybe this makes some sense, but I would rather put my time into listening than into my own tweaking of the sound. Not every recording will be mastered to my preference, but then not every movie or TV show is going to have the lighting and cinematography that I prefer. Unless it is so bad that I can't enjoy it I really don't feel the need to remaster it myself, and I can always adjust equalization if it is that bad.